Sitemap

QCode.cc homepage — learn about our developer subscription service for Claude and Codex

View Claude Code API pricing plans

Customized Claude Code solutions for enterprise teams

Become a QCode Channel Partner — branded domain + one key for three CLIs + 95-75 tier discount

View pricing details, service rates and popularity for all supported AI models

Complete Claude Code documentation and technical guides

Learn the basic concepts and core features of Claude Code

Claude Code CLI tool installation and environment setup

Get started with Claude Code in 5 minutes

Detailed guide to Model Context Protocol extensions

Common error codes and their solutions

Frequently asked questions about Claude Code

Claude Code best practices guide

Codex CLI integration guide

AI-powered documentation assistant

Detailed Claude Code CLI tool installation, configuration and usage tutorial

Common questions and technical support for Claude Code usage

In-depth analysis of Claude Code's unique advantages and competitiveness

Comprehensive comparison of features, pricing, and use cases

Claude Code MCP ecosystem and configuration tutorial

Claude Code Hooks system and CI/CD automation guide

Project memory file writing and team collaboration standards

Claude Code token savings and smart routing strategies

Complete guide for Chinese developers to use Claude Code, with Alipay support

Claude Code Voice Mode setup and workflow best practices

Track the leaked Mythos / Capybara model tier based on public reports and draft materials

Complete guide to Anthropic's fully managed cloud AI agent platform

Complete developer guide for Claude Agent SDK

Claude Code plugin ecosystem and custom plugin tutorial

Claude Cowork desktop complete guide

Multi-agent collaboration complete guide

Orchestrate hundreds of subagents — triggers, /deep-research, limits, and cost.

Claude Code effort levels — high, xhigh, max — and what ultracode adds.

Anthropic's next flagship Mythos: official preview, GA outlook and QCode access

Programmatically query API key expiry and usage with a read-only token

Why Claude accounts get banned and the compliant, API-based path to keep using Claude models.

0731 re-post-train: agent scores, $0.14/$0.28, local weights

1.6T-class GA, agent benchmarks, peak/off-peak pricing, QCode catalog price

AA Intelligence 52 vs 55. Cost/task $0.06 vs $1.54. Speed 115 vs 69 tok/s. Both 1M context. Flash is open; Sonnet is the Claude default.

Same HF table: TB2.1 87.9 vs 88.0, DeepSWE 62.7 vs 70.0, HLE w/tools 60.0 vs 63.0. Fable still leads most rows; Pro wins price and MIT weights.

Open 1.7T agent flagship vs the closed GPT-5.6 top tier: 53 vs 61 on intelligence, $0.66/$1.98 off-peak input/output ($1.32/$3.96 peak; official pricing page 2026-09-20) vs $4/$20 (Sol promo price, at least through 2026-11-21).

Grok 4.6 shipped 2026-08-12. AA 61 (tie with GPT-5.6 Sol), $2/$6, 500K context. Strong long agents; DeepSWE 65.9% still behind Sol/Fable.

Opus 5 still leads AA (63 vs 61) and DeepSWE. Grok 4.6 is ~1/4 the output price and ~half the Briefcase turns (53 vs 103).

Grok 4.7 in the catalog: specs, timeline and today's alternatives

Pre-layout comparison: live 4.6 vs Fable 5 numbers, no fake 4.7 benches

As of 2026-10-07 Grok Bot comes with every paid individual Cursor plan and Cursor Teams, runs on macOS, Windows, Linux, iOS, Android; bots share a cloud PC.

Grok Bot cloud computer RAM and disk: unpublished officially; early users report Debian, 8 vCPU, 16GB RAM and ~128GB disk.

As of 2026-10-07: OpenCode Go is $10/mo, Go Plus $40/mo, no $5 first month listed. Per-model dollar limits: 5h = 20% of monthly, week = 50%. Go ≠ Zen.

Official X: new V4 Flash is in OpenCode Go and Free. Ids opencode-go/deepseek-v4-flash and deepseek-v4-pro. Flash is the volume lane; Pro burns the dollar cap faster. DeepSeek ZDR: as of the 2026-08-30 fetch, official text still said valid through August 31, 2026.

As of 2026-10-07, OpenCode's Go FAQ gives free models (Big Pickle plus promos) 200 requests/day. No Go Free plan is listed: Go is $10/mo, Go Plus $40/mo.

Official DeepSeek agent harness (dsh). Everything is a plugin on the Cordis kernel. npx @deepseek-ai/dsh web. MIT, breaking changes. Minimal mode is how Flash 0731 / Pro 0813 official benches were run.

Kimi Work is Moonshot’s local desktop agent for knowledge work: folders, WebBridge browser automation, Cron, Agent Swarm, finance data. Windows + Apple silicon. K3 needs app ≥ 3.1.0.

Opus 5 leads most independent aggregates (~83 vs ~80) with overlapping intervals. K3 is 2.8T, 1M, native vision, $3/$15, open weights. Official K3 blog says it still trails Fable 5 and Sol.

Shipped 14 Aug 2026; tracking API and weights; not listed on QCode

No Gemini 4 or Gemini 4 Pro SKU yet — pre-training confirmed, 3.5 Pro delayed

There is no official DeepSeek Coding Plan / Code Plan SKU. Consumer chat is free; money is prepaid API (peak/off-peak from 2026-08-16). The query maps to: the 2026 coding roadmap, a wish for token-metered prepay, or Harness plan mode.

13 Aug 2026 GA: intro $0.75/$3.75, 1M context, coding benches

14 Aug 2026 launch: same-base post-training, Coding Plan, staged weights

Intro $0.75/$3.75 through year-end, then doubles; Batch/Flex half

Same intro price, coding benchmark comparison; both ids in the catalog

2.4T MoE, $2/$6, open weights; on QCode (catalogue checked 2026-09-20)

27B native multimodal, 262K, YaRN 1M, Apache 2.0; not on QCode

id qwen3.7-max, 1M, $2.50/$7.50; retired 2026-09-09, superseded by 3.8-Max

750 tok/s, vendor 14×; OpenAI limited preview, not a QCode tier

Text watermark + C2PA: a mark ≠ author, no mark ≠ not AI

kimi-k3: 2.8T, 1M, $3/$0.30/$15, thinking will not turn off

Intro rate vs $2/$10; both listed, same endpoint

Same-base post-training; 5.2 listed, 5.3 not stocked

Facts, impact and practical developer options behind the $7B+ deal.

Judge a new model against your own work instead of public leaderboards.

A billing-definition checklist and a predictable-cost alternative.

How 529 differs from 429/503, correct backoff, and multi-route availability.

Reasoning-tier cost differences, the reality of extra credits, and alternatives.

5-hour window vs weekly cap, the +50% boost that ran until 2026-09-13, and what to do when quota is gone.

The vetted offensive model: who can apply, pricing, and how it differs from Sol

Three boards, three winners: the official 83.8 / AA 89.5 / vals 85.77 gap fully explained

Three fields to QCode; choosing between BYOK pay-per-use and a $20 subscription

Five-surface architecture, model list, $19.99/$100 pricing, and the boundaries of a closed platform

A parallel multi-agent workspace, $4B ARR, and the SpaceX $60B acquisition timeline

Four at $20, three at $100: after convergence, how to choose between subscription and pay-per-use

Orchestration/execution/review layering, per-task model routing, one key as the base

The header, supported models, cache behavior and forwarding pitfalls

80.3 / 80.0 / the official board: the truth about data splits and scaffolds

Training started (confirmed by multiple outlets); release date and specs all unknown

The September rumor has no official source; V4's price hike and in-stock status are facts

Peak/off-peak pricing details, the real increase, and off-peak scheduling to save

The three-owner timeline, $20/mo pricing, and migration advice

A half-point war: what to compare once benchmarks tie

Open flagship vs closed flagship: a 1.2-point gap and an order-of-magnitude price gap

Two routes to the open-source lead: post-training vs giant base

Console prepaid credits, auto-reload, payment methods, one-year expiry and Max/Team monthly credits, compared with QCode top-ups.

4.1.22 started sending the fallbacks option and custom endpoints returned 400; 4.1.23 (2026-10-07) fixes it. Plus the QCode Anthropic-provider setup.

Which model replaces each of the three, how to handle snapshots and default effort, and what changes in Codex (official table checked 2026-10-08)

Five tiers became three on 2026-10-06: $5 / $100 / $500 thresholds, $500 / $5,000 / $200,000 monthly limits, with RPM and TPM.

Which models cannot turn thinking off, the five effort levels and their defaults, where budget_tokens stands, and Claude Code settings. Checked 2026-10-08.

Loaded at startup, replaces the bundled catalog, profile wins; the 0.160.0 explicit provider catalog change and /model troubleshooting.

Live since 2026-10-07: $100 / $200 / up to $500, claimed into a Console organization, not for Claude Code, no rollover.

Added for ZDR orgs in 2025-04, removed by PR #3212 on 2025-09-05; not in the current config-reference, and old configs get a warning.

The five breaking changes from Haiku 4.5 to Haiku 5.5, each with the official error behaviour and fix.

Two price tiers ($0.10 / $0.50 up to 100K, $0.50 / $2.50 above), 1M / 128K, platform IDs and how it differs from Haiku 4.5.

As of 2026-10-08: local CLI, MCP, subagents, auto and 1M work; Remote Control, cloud sessions and other account-side features don't.

Which ID replaces each of the six GPT-5 / o3 snapshots, what to do with aliases, and a migration checklist (official table checked 2026-10-08)

No new official change found as of 2026-10-08: 5-hour sessions, fixed weekly reset time and your options after hitting the limit.

The advisor_20260301 400, Extra inputs are not permitted, rejected beta headers and 1M context: exact error text, cause and fixed version.

codex login, switching to an API key, auth.json, env_key vs requires_openai_auth, and 401 checks for QCode.

16 BYOK tools: protocol, config location, Claude support and setup page, checked 2026-10-08.

Payment methods (card, Apple Pay, Google Pay, crypto), Japanese docs and ANTHROPIC_BASE_URL setup steps.

The three official install commands, the QCode one-click setup script, verification and common install errors.

A plan is not an API, but Max and Team now include monthly API credits (not for Claude Code, not on Pro): monthly fees, five-hour and weekly limits, Claude Code auth precedence, and when to pick a per-token API key.

Zen is pay-as-you-go, Go is $10 or $40 a month: prices, models, limits and how to choose, checked 2026-10-07.

One cr_ key for DeepSeek, GLM, Kimi and Qwen in Claude Code: same address, switch with /model.

Configure a Gateway in developer mode: Anthropic Messages API only, Claude models only, no ANTHROPIC_BASE_URL, plus the Claude Code differences.

Credentials in claudeCode.environmentVariables, settings.json shared with the CLI, plus login prompt and model selection fixes.

Where keys go in Settings → Models, which slot can change the Base URL, why Tab skips your key, and the QCode setup.

BYOK setup for VS Code Custom Endpoint and Copilot CLI: URL forms, plan requirements and the JetBrains limit.

Pro 200 keeps the old allowance through 2026-10-29; Pro 500 and Ultrafast at 8x / 6x.

Interactive sessions now default to auto mode (docs say 2.1.283, changelog 2.1.284): the six permission modes, how to turn it off, and notes for teams and CI.

Custom provider: Claude via Anthropic Messages at /api, GPT and Chinese models via OpenAI Compatible at /openai/v1, plus common errors.

Install, add a QCode configuration, switch between Claude Code and Codex in one click, and fix switches that don't take effect or 401 errors.

Cache read and write prices, TTLs and minimum cache lengths for Claude, GPT and DeepSeek, with two recomputable examples (as of 2026-10-07).

Sonnet 4.5 retirement, GPT-5.6 Sol promo pricing, OpenAI shutdowns and three conferences, each with its source and check date.

How to fill in a Trae custom model: API format, request URL, model ID and API key, plus the Auto mode limit and common errors.

One sub-API key per person, a daily quota per key, qot_ usage queries, offboarding, and how Anthropic Team / Enterprise compare.

How to set model_provider, base_url and env_key in config.toml, plus common errors and model picks (checked 2026-10-07).

Released 2026-09-29: $2 / $10, cached input $0.10, 1.05M context, full-request pricing above 272K, compared line by line with gpt-6-sol.

Long-context billing rules and three worked examples for Claude, GPT-6 Sol / 6.1 Sol, Grok 4.7 and DeepSeek V4.1 Flash (as of 2026-10-07).

From 2.1.285, a custom base URL gets 1M automatically; use /autocompact 200k if your gateway stops at 200K, plus hint headers and 529 backoff.

Claude via the Anthropic type at /api, GPT and Chinese models via the OpenAI type at /openai, one cr_ key, plus common errors.

Apple Pay, Google Pay, cards, crypto, Alipay and WeChat Pay: currency, fee labels, amount range and the three top-up steps.

Each of the five official breaking changes paired with its error string, plus the between_tools and effort combination limits and the response-shape change.

Items where a vendor gave a date or a hedge, each with its source and the date it was checked.

Official tables side by side, how each vendor handles long context, five recomputable examples, and the 2026-09-29 release of gpt-6.1-sol as background.

OpenAI's own sentences from 2026-09-29, item by item: the id, pricing and specification of gpt-6.1-sol, three API changes, dots and Pro 500, plus the fields OpenAI has still not published.

Released 2026-10-07: model ID, both price rows, 1M / 128K, and where Haiku 4.5 stands.

Official rates, billing rules and recomputable examples for two OpenAI reasoning models (as of 2026-09-29).

Official wording for the concept, plus Atlas' access limits and the AMD agreement.

Documented endpoint values, what ANTHROPIC_BASE_URL changes, and how to measure responsiveness yourself — without publishing numbers.

Choosing an AI coding stack as a DACH team: endpoint freedom, workflow shape, instruction files — no price ranking.

CLAUDE.md and AGENTS.md for German replies: what is documented, what is not, and a paste-ready rule block.

Cards, wallets, crypto and the corporate bank transfer path — what QCode documents for European buyers.

Availability, endpoints, payment and documentation language for Claude Code and Codex from Germany, Austria and Switzerland.

Same context window, different unit prices: $2/$10 versus $4/$20, cache reads at $0.20, with worked examples.

Released 2026-09-28: 1M / 128K, $2 / $10, cache reads $0.10 since 2026-10-07, plus default effort and the five breaking changes.

The RSI definition and each lab's 2026-09 wording, marked for self-report and audit

What the reporting says: the quotes on record, the two regressions, the timeline and the models still callable today.

Claude Opus 5.5 guide: pricing, caching, the comparison with Opus 5, calling it on QCode

Headers, block placement, the 400 and the billing wording for both paths (official docs captured 2026-09-21)

The 9 old ids, their substitutes, the line to edit and how to verify (vendor table re-fetched 2026-10-07)

The date, all six mappings, the four models already removed on 2026-10-02, admin auto-enable terms and verification steps (checked 2026-10-07)

Retirement date, empty replacement column, 410 Gone and the verification order (checked 2026-09-20)

Removal date, replacement id, the field to edit and how to verify (work to 2026-10-01)

Earliest shutdown 2026-09-30, the replacement id, which line changes (table archived 2026-09-21, catalog checked 2026-09-20)

The official table and cost claims, what is missing, and our availability check (2026-09-20)

AWS prices and explicit-cache parameters, and how Bedrock compares to other routes (docs checked 2026-09-20)

What TypeSafe's Jev is, which claims the vendor evidences, and how to do the same work without it on QCode (checked 2026-09-20)

Officially dated, no successor id.

Official table (checked 2026-10-07): gpt-4-turbo and gpt-4 to gpt-5.6-sol, gpt-3.5-turbo to gpt-5.6-terra.

Replacements and steps for the 2026-09-28 and 2026-10-23 batches (checked 2026-10-07)

Official, rumor, unpublished.

A post is not the table. Switch to ids with usage.

Two in-sale workhorses, not a “what is X” page.

Status page, policy, quota.

Voice per minute, backend per token — two invoices.

503 / 429 / quota, and how to use status.openai.com.

As of 2026-10-07 Pro has $100 / $200 / $500 tiers and Pro 200 is open again; what to check if it still fails, and when to use API keys.

Two flagships at identical list prices: the differences are thinking mode, latency tier and knowledge cutoff.

Identical windows, 2.5x price gap; the difference is the none reasoning tier and the knowledge cutoff.

Pre-training officially confirmed to have started; no date or specs published, and 3.5 Pro is still queued ahead.

Officially confirmed to exist and in training, with no name or date published. Official wording, timeline and migration prep.

The four hard constraints in the official migration guide, and the one that gets mistaken for a bug.

Official intro price $0.75/$3.75 through 2026-12-31, then double; the Cyber variant is limited to the Fairwind Program.

Official 2026-09-10 release: new id, 1M/384K, peak/off-peak pricing, routing conflict, and which id to use on QCode.

The new id since 09-10, temporary routing of legacy flash ids, vision-exp retirement, the v4-pro routing saga, and which id to set on QCode.

Official says Flash ahead on several axes but V4 Pro continues; price/context/choice and QCode comparison.

All six rates, peak windows in local timezones, the gap to V4 Pro and concurrency limits.

Official scope, how to tell if your Codex is affected, and the official per-plan replacements gpt-6-sol and gpt-6-luna.

How it differs from quota and 429, the user-filed issue record, and the practical model-switch and backoff moves.

From community label to official model: release date, gpt-6-sol specs and pricing, and how to use it on QCode.

What the official table says as of 2026-10-08, the 5.1 release date, the new 5.5 models, and today's callable id.

What Anthropic released on 2026-09-22 is Opus 5.5; 5.1 / 5.2 unannounced; the id that works today.

The single official sentence, the same-day changelog line beside it, and which id to set today.

Official credits table: Astra is exactly 2.5× GPT-5.6 Sol. Per-plan message allowances and three ways to burn less.

The two environment variables, the settings.json form, apiKeyHelper, and the documented web-session limitation.

Cache reads 75% cheaper, input and output unchanged, 1M context, and how to decide whether to switch from Fable 5.

Official promotion dates, how the five-hour window differs, and a /usage self-check plus downgrade path.

Three reset times, four reasons a bill will not reconcile, and the refund and transfer terms.

Data classification, five contract questions, and the line between verifiable and contractual.

Pooled versus daily-cap quota, per-account concurrency and headcount, and sub-key attribution.

Per-transaction caps, who absorbs the fee and settlement time across four channels.

Four self-tests: multiplier conversion, cache hits, echoed model and usage reconciliation.

The post-hike open-flagship math: DeepSeek off-peak, GLM at peak

The real deployment cost and break-even point of 2.8T open weights

What the confidential filing means, the impact on API users, and hedging strategies

The 30-day review window, the voluntary framework, the open-weights exemption, and the effect on release cadence

TB 3.0 vs. 2.1, current rankings, and which leaders are sellable

The 10 evaluations and category weights as of 2026-10-07, compared with the old v4.1 weights

SWE-bench score, pricing, and how to use it

Pricing, positioning, and real call-volume data on QCode

Positioning, use cases, and current QCode status

Delisted from QCode on 2026-09-09; use kimi-k3 from the same family

On sale alongside Max, real usage data verified

Official openai_compatible: api_url, model id, key

OpenAI Chat Completions, public URL, do not pick Auto

Official OpenAI-compatible: URL, API key, Test Connection

Stable id, 1M context, Standard $1.50/$9.00

Stable id, Standard $0.30/$2.50, 1M context

OpenAI Codex CLI installation, configuration, and comparison

OpenAI's strongest agentic coding model, plus Goal Mode and May 2026 Codex features.

Deep dive into OpenAI's latest GPT-5.4 and GPT-5.4-Codex model capabilities

Codex desktop app and ChatGPT Super App complete guide

2026 flagship model benchmark — GPT-5.5 vs Claude 4.7 across all dimensions + QCode dual API access

2026 flagship model benchmark — GPT-5.5 vs Claude 4.8 across all dimensions + QCode dual API access

Sharper, more reliable judgement on agentic tasks, 4x fewer code flaws than Opus 4.7, Online-Mind2Web 84%, native 1M context — at the same price as Opus 4.7 ($5/$25 per MTok)

Anthropic's most capable model yet: native 1M context, 128K output, $10/$50 per MTok, migration notes and QCode access

The new flagship at the same price and window, with a 512-token cache floor

Release, specs, pricing and how to access Claude Sonnet 5 (claude-sonnet-5) — 1M context, agentic coding, close to Opus 4.8 at a lower price.

The standard $2/$10 per MTok rate after the September increase was cancelled — plus cache tiers and the new-tokenizer cost caveat.

Price, coding and which Claude model to use — Sonnet 5 as the value default, Opus 4.8 for the hardest work.

Drop-in model-id swap plus three breaking changes and the new tokenizer — for API and relay users. Sonnet 4.6 stays available.

Release date and a plan-by-plan, provider-by-provider availability tracker for Claude Sonnet 5, including access via QCode.

Production security practices, sandboxing and permission controls for AI Agent MCP servers

Decision framework and hands-on patterns for routing tasks across Claude, GPT and Chinese models

2026 开发者如何高效 review AI Agent 生成的代码与 PR

1M+ 上下文窗口下的实用 coding 技巧与缓存策略

基准分数如何转化为真实产品交付速度与可靠性

Model id claude-sonnet-5, omitting sampling params, the effort parameter and re-tuning max_tokens — with QCode relay access.

An honest, source-labeled comparison — why there is no shared benchmark; Claude and GPT switch on one QCode key, while QCode does not currently offer the Gemini series.

A dated status tracker for Claude Fable 5 — available on the API throughout, tiered by plan on Anthropic subscriptions since 2026-07-19.

GPT-5.6 tier benchmarks, pricing and access — now GA and live on QCode

The released Opus 4.8 vs the rumored GPT-5.6, an honest side-by-side + QCode dual-API access

How to choose between the three tiers: benchmarks, prices and scenario guide

Four steps to run gpt-5.6-sol / terra in Codex CLI via QCode

Official Sol / Terra / Luna prices, cache billing and cost estimates

OpenAI's 2026 flagship image generation model — capabilities, pricing ($0.30 per image), QCode access

A calm, factual look at why Codex and ChatGPT accounts get suspended and why an API-based path is more stable.

The 5-hour rolling limit returned on 2026-07-30 — timeline, community reports and what to do when you hit it.

Shipped 2026-09-03. Specs, pricing, four migration constraints, and three things OpenAI still has not published.

GPT-5.7 never shipped; the candidate name resolved to GPT-6 Astra on 2026-09-03. Includes the 1.5M context debunk.

Luna $0.20/$1.20 after the July 30 cut, with Terra and Sol rates and Codex notes

Decision framework + three-camp overview + navigation to this site's full lineup content.

Where the 80.3% figure comes from, official methodology vs. independent reproduction, and how much it should weigh in your decision

Official API rate breakdown, flagship price comparison, and estimated task costs

Same architecture, same benchmark score; the availability and safety-routing differences, plus evidence Mythos 5 has no public pricing

Price, context window, and benchmark methodology compared, plus how to call both on QCode

The 8/31 shutdown timeline, who's affected, and how to move to Terra/Luna

July 21 release specs, $1.50/$7.50 pricing, and cost context for coding APIs

Office multi-agent overview, Custom API setup angle, and coding-tool boundary

BTC mainnet plus four kBTC networks and ten ETH networks, zero platform fee, with kBTC explained.

Crypto tops up an ordinary USD balance that reaches every model on offer, with no payment-method tiering.

18 USDC networks, zero platform fee, with the Canton memo and Stellar muxed address explained.

18 USDT networks, zero platform fee, with per-chain cost notes and the USDT0 pitfall explained.

Nine coins, 56 networks, zero platform fee for AI coding API credits.

Detailed comparison of mainstream AI coding tools: Claude Code vs Cursor vs Gemini CLI

Complete AI programming tutorial from beginner to advanced, including prompt engineering

Professional AI programming tips, prompt engineering and workflow optimization guide

Success stories and experience sharing of Claude Code applications in real projects

2025's most authoritative in-depth review and selection advice for AI coding tools

Latest developments in AI coding tools for 2026: Claude 4.6, Agent Teams, Codex Desktop and more

The paradigm shift from conversational to autonomous AI coding

Complete guide to AI multi-agent programming

7-tier plans, 1:1 rate, token estimates, no window limits, booster packs

A neutral overview of Claude and Codex subscription bans and why pay-as-you-go API access is more stable for coding.

Complete guide to the OpenClaw open-source AI Agent framework architecture and features

Guide to OpenClaw 5700+ AI agent skills ecosystem

Complete guide to OpenCode open-source AI coding assistant

Git-native AI coding assistant with Claude model integration

New paradigm for AI agent system design, three-generation evolution from Prompt/Context/Harness

Natural language-driven AI programming, complete guide to 2026's hottest coding method

AI agent information architecture design, the new paradigm beyond Prompt Engineering

Live release and availability status of every 2026 frontier model.

When Gemini 3.5 Pro launches and the Gemini models available today.

The strongest open-weight coding model of 2026 — benchmarks and caveats.

Full Apache 2.0 release: self-compile, attach your own inference, privacy timeline, hybrid patterns.

xAI's flagship: pricing, Claude/GPT comparisons, access and quota hedging.

Kimi K3 soft launch status: confirmed vs community rumors.

Claude Code weekly/rolling quota windows and reset rules explained.

Mid-2026 Codex quota anomaly, official reset and banked mechanics.

320B-A18B, 1M context, MIT weights; promo ends 9/9; real DeepSWE 63.4%

50% off $0.075/$0.25 to 9/9; why cache reads are the real cost driver

How to tell apart three error types and what to do about each

Why /status shows headroom but requests get rejected, and a 3-step way to tell why

Cursor Models pool (Grok 4.7/4.6/4.5, Composer 2.5) vs Other Models pool and the reset rules, checked 2026-10-07

Telling apart a retired sign-in path from a quota issue, and migration options

The three disguises of context overflow and how to tell them apart

Four causes of stream cutoffs and a hop-by-hop triage method

The window's starting point and how it stacks with the weekly limit

Weekly-limit rules; the +50% promo ended 2026-09-13, standard +25%

The token-resend mechanism, usage estimation, and reduction methods

How to calculate the break-even point

How to compliantly orchestrate quota across Claude Code, Codex, and Cursor

СБП, Russian cards, and crypto — the three payment paths explained

How to diagnose the error and Russia payment methods explained

A 429 with no retry-after is the org monthly spend cap, not RPM throttling

Under coding originators Sol is 272000, not the 1.05M model spec

The literal means prepaid usage credits are 0, not weekly limit or the monthly cap

Notice sent 2026-09-30; Copilot dropped it on 09-01 — migrate the two calendars separately

Artboards in CLI/Desktop before implementation; research preview, not GA

Desktop resumes after official refill; does not cancel the 5-hour window or weekly limit

Copilot-wide 09-01 drop of the listed models; not the same day as Anthropic API retirement

Released 2026-08-26 and its public list price; qwen3.8-flash is in our catalog and usage table (checked 2026-09-18)

xhigh/max plus thinking off is 400; 2.1.251 clients send high

Two sold Flash tiers: official price, 1M window, MIT; same-endpoint switch

2.1.251: official hooks before a model switch, block / confirm / annotate

Paid-plan early beta from 2026-08-17; GitHub-free cold start on 08-27

What Orca is and is not, parallel git worktrees, which CLI agents it runs, install paths and the name clash (official README and docs checked 2026-09-22)

How the 5-hour and weekly limits stack up, N agents each carrying their own context, and the cache price gap (docs and price list checked 2026-09-22)

Endpoint fields and config locations for 11 CLI agents; undocumented rows marked as such (checked 2026-09-22)

CLI command name, the Homebrew command with the tap, picking from three Linux packages, the CLI missing its command, official docs on SSH and phone pairing (checked 2026-09-23)

15 tools side by side on the 2026-09-22 stars, licences, activity and form; the form and model-source columns only carry what each README states word for word. Checked 2026-09-23.

Learn how we protect your privacy and data security

Terms and conditions for using QCode.cc services