Skip to content

About

📡 Legal free LLM APIs & coding agents — self-updating, probe-verified twice a week. Free tiers, no-card trials, free models.

mvalentsev.github.io/awesome-free-ai-coding/

Topics

ai-codingawesomeawesome-listclaudeclaude-codecoding-agentsdeveloper-toolsfree-aifree-apifree-llmfree-llm-apifree-modelsfree-tiergemini-apigptllmllm-apino-credit-cardopenai-compatibleopenrouter

Resources

Readme
MIT license

Contributing

Contributing
Activity

Stars

128 stars

Watchers

4 watching

Forks

26 forks
Report repository
BranchesTags

Latest commit

 

History

707 Commits

Folders and files

NameName
Last commit message
Last commit date
.githooks
.githooks
 
 
.github
.github
 
 
_layouts
_layouts
 
 
assets
assets
 
 
configs
configs
 
 
models
models
 
 
providers
providers
 
 
src/freetier_radar
src/freetier_radar
 
 
templates
templates
 
 
tests
tests
 
 
.gitignore
.gitignore
 
 
.imgbotconfig
.imgbotconfig
 
 
404.html
404.html
 
 
AGENTS.md
AGENTS.md
 
 
CONTRIBUTING.md
CONTRIBUTING.md
 
 
LICENSE
LICENSE
 
 
README.md
README.md
 
 
_config.yml
_config.yml
 
 
blocklist.yaml
blocklist.yaml
 
 
browse.html
browse.html
 
 
dismissed.yaml
dismissed.yaml
 
 
eb68c254f1e03877b906ccc800002691.txt
eb68c254f1e03877b906ccc800002691.txt
 
 
feed.xml
feed.xml
 
 
history.jsonl
history.jsonl
 
 
index.html
index.html
 
 
index.json
index.json
 
 
llms.txt
llms.txt
 
 
pyproject.toml
pyproject.toml
 
 
registry.yaml
registry.yaml
 
 
sources.yaml
sources.yaml
 
 
uv.lock
uv.lock
 
 
watchlist.yaml
watchlist.yaml
 
 

Repository files navigation

Live probes check every row against the vendor's API or pricing page twice a week. Ended offers move to the Archive; verified dates link to the evidence.

🆕 New this week: Merkium · Mixedbread · Codacy · DeepSource · Graphite · Liquid Inference · LLM.kiwi · Sourcery · Summer Engine · NoTrack API · Kiosapi.id · 1min.AI · Infron · Zro · CodeAnt AI · CodeRabbit · cubic · Greptile

🚀 Start here

Free is not the same as weak. Coding agents on a $0 plan, no card required:

  • opencode — big-pickle · nemotron-3-ultra · nemotron-3.5-lightning · mimo-v2.6-flash · +3 more
  • Kilo Code — nemotron-3-ultra · north-mini-code · nemotron-3.5-lightning · dots-3-note · +4 more
  • Kiro — claude-sonnet-4.5 · claude-sonnet-4 · glm-5 · minimax-m2.5 · +3 more
  • Google Antigravity — gemini-3.8-flash · claude-opus-4.6 · claude-sonnet-4.6 · gpt-oss-120b

Free models, compared. Current Artificial Analysis Intelligence Index v5.0.1 scores, read 2026-11-06, highest first. Each linked provider offers the model free under the conditions beside its name; 💳 means a card is required.

Where each one is free — every live offer beside its model, in the chart's order

All 130 free model families, including those without current scores, remain in the model index.

Or pick by what you need:

The top of each section in the list's own order — how rows are ordered; the Claude Code line names gateways with an Anthropic-format route every run calls, one shell function each in configs/claude-code.sh; the Codex CLI line names lanes Codex calls directly — ones that took its request with no key, which every run sends again, and ones whose vendor's own page sets Codex up, which every run reads back — one profile each in configs/codex/.

Try without an account: Kilo Code — a rate-limited connectivity check:

curl -s https://api.kilo.ai/api/gateway/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model":"openrouter/free","messages":[{"role":"user","content":"2+2? MAKE NO MISTAKES."}]}'

Use free IDs without a key; metered IDs require sign-in. kilo-auto/free and openrouter/free choose a free model automatically. LiquidAI advises against agentic coding with lfm-2.5-2.6b.

Connect your coding agent: base URLs, keys and client configs.

📋 The list

💳 — the row wants a card on file, 4 of 103, and no other row asks for one · 👁 — what you send may be used to train models, in the vendor's own words · 🧪 — new, provisional for two weeks of probes · verified — the last day a live probe confirmed the offer; it opens the row's page: the quota in the vendor's words, the evidence, the history

🤖 Coding agents & CLIs

27 live · not one of them asks for a card · sorted by how much work you can get done for free

🔌 LLM APIs with free tier

35 live · 33 of them ask for no card · sorted by how much work you can get done for free

🎁 Trials (no card when possible)

23 live · not one of them asks for a card · sorted by how much work you can get done for free

🧭 Aggregators (one key, many providers)

18 live · 16 of them ask for no card · sorted by how much work you can get done for free

🧠 Looking for one model in particular? The website's model index names all 130 model families on the list and everyone who serves each one free; the comparison is under Start here. A model gets a page of its own once two rows or more serve it free, or it measures notable, strong or frontier, or it has a dated free promotion.

🕰 What changed — the last 10 registry events, and an Atom feed of each new one
When What Details
2026-11-06 🔄 Free models changed AIHubMix (free models) added gemma-4-26b-a4b, gemma-4-31b
2026-11-06 🔄 Free models changed LLMTR dropped jais-3-mova-36b-a4b, k2-horizon-375b-a23b
2026-11-05 🔄 Free models changed NovAI added glm-4.1v-thinking-flash, glm-4.6v-flash, glm-4.7-flash
2026-11-05 🔄 Free models changed NVIDIA NIM (build.nvidia.com) added nemotron-3.5-bolt-e4b
2026-11-05 ➕ Added Merkium Independent model developer with a free OpenAI-compatible API and chat service; current API models have no per-request or per-token charge
2026-11-05 🔄 Free models changed Cline added beam-1.0
2026-11-05 🔄 Free models changed Cline added nemotron-3.5-lightning
2026-11-04 ➕ Added Mixedbread OpenAI-compatible API for the Toast 1 knowledge and search model, with $5 in one-time Starter credits and no card required
2026-11-04 ➕ Added Codacy AI pull-request reviews for public open-source GitHub repositories, with a separate no-card trial for private repositories
2026-11-03 ➕ Added DeepSource AI code review and suggested fixes on existing repositories, with a no-card introductory trial and bundled AI Review credits

Arrivals, departures and free-model changes. Full log: history.jsonl; updates: Atom feed.

📦 Archive

23 offers and why they left
Tool Why it left
LLM Tech vendor-announced shutdown on 2026-11-01
Haimaker documented free promotions ended on 2026-11-01
Poolside Platform vendor-announced shutdown on 2026-10-31
Amazon Q Developer delisted on 2026-10-26: closed to new Free accounts and new subscriptions; existing IDE users retain access until the announced end of support
uncloseai (unturf) delisted on 2026-10-22: public beta closed; the endpoint requires a key and key registration is closed
Yolo-Auto delisted on 2026-10-15: free accounts now include 0 requests/week and API inference requires a paid plan
Crush + Charm Hyper delisted on 2026-10-15: The current FAQ retires the monthly free Hypercredit allowance; access requires a paid plan or prepaid credits from $5 (https://hyper.charm.land/faq).
Blue Claw Network delisted on 2026-10-15: Current onboarding requires purchasing prepaid credits before API use; the previous $5 welcome grant is no longer advertised (https://blueclaw.network/quickstart).
Hugging Face Inference Providers delisted on 2026-10-08: free accounts no longer receive inference credits — the current billing table lists None and requires purchasing credits or a paid subscription
SambaNova Cloud delisted on 2026-09-23: the console requires a payment method and purchased credits for first requests, while documentation still describes a no-card Free tier; obtainable new-account Free activation remains unconfirmed
Scaleway Generative APIs delisted on 2026-09-16: the documented Free token allowance has unresolved renewal and new-account activation conditions; generic resource-ordering and standard-rate payment requirements do not establish every Free Serverless mode
Kenari delisted on 2026-09-16: rejected for cause — the operator's own JavaScript bundle showed its capacity coming from pooled ChatGPT and Codex OAuth credentials, captcha solvers and a proxy pool that multiplies per-IP free quotas
Infomaniak AI Services delisted on 2026-09-16: a one-month wallet of a million credits, and "A credit card is required to start using the API" — a one-off credit behind a card, which CONTRIBUTING does not admit
Cerebras Inference delisted on 2026-09-16: its only free offer was $5 of credit granted after a verified payment method is added, expiring in 30 days — a one-off credit behind a card, which CONTRIBUTING does not admit
LongCat API Platform delisted on 2026-08-14: the daily free quota was gone before the row was listed — the change log retired the Flash line on 2026-05-29 and switched billing on for LongCat-2.0 on 2026-06-30, and no LongCat page still published the 100K tokens a day
Reka AI delisted on 2026-08-11: the $10 of free credits a month rested on one dated announcement post; no live Reka page corroborated a recurring grant, and the platform sells pay-as-you-go credits
AI21 Labs (Jamba) vendor-announced shutdown on 2026-08-09
GitHub Models vendor-announced shutdown on 2026-07-30
DeepSeek Platform delisted on 2026-07-27: the 5M-token signup grant was only ever reported by third parties, and DeepSeek's own pricing page prices every model and names no free grant
MiMo Code vendor-announced shutdown on 2026-07-26
Puter.js (free LLM API) delisted on 2026-07-19: the recurring Free SDK and HTTP-driver allowance is separate from the OpenAI- and Anthropic-compatible coding-agent endpoints, which require a paid plan; a supported Free coding-client integration is not established
Easy GonkaAI API delisted on 2026-07-19: rejected for cause — its /for-agents page carried a prompt injection aimed at AI agents (auto-signup, credential exfiltration, referral spam)
Aider delisted on 2026-07-19: BYOK only — the tool is free and open source but bundles no model access of its own, and its docs send readers to other vendors' free tiers, which this list carries directly

Provider pages retain the offer, evidence and history. A row archived by its probe returns when it passes again. Merged entries keep their old pages, pointing to the retained row. See the archive rules below.

🔭 Checked and not listed — 276 services whose free tier this list could not find or could not verify, each with the reason on the date it was read and what would change the answer: the whole list of them. Nothing there is disqualified, and every verdict expires after 90 days; domains rejected for cause are a separate file, blocklist.yaml.

🔧 Plug it into your agent

Connection table — base URLs, keys and client notes for 69 live OpenAI-compatible APIs, plus available Claude Code routes and direct Codex profiles. Copy buttons on the website. These configs are regenerated on every update:

  • configs/opencode.json — Drop-in opencode config with every provider wired up: keys via {env:...}, keyless endpoints work immediately
  • configs/litellm.yaml — LiteLLM proxy config: env -u OPENAI_API_KEY LITELLM_DANGEROUSLY_PERMIT_WEAK_OR_UNSET_MASTER_KEY=true litellm --config configs/litellm.yaml --host 127.0.0.1 puts every free model LiteLLM can call behind one local endpoint; ask for free/nokey instead of a model and a call moves to the next free lane when one runs out — LiteLLM would send your OPENAI_API_KEY to a lane whose own key is not set
  • configs/claude-code.sh — One shell function per gateway that serves the Anthropic Messages format: source it and run Claude Code on a free lane, key and model filled in where the row lists one
  • configs/codex/litellm.config.toml — Codex CLI profile over the LiteLLM config: Codex speaks only the OpenAI Responses API, and the proxy answers it for every lane in that file from the lane's chat completions — copy it to ~/.codex/ and run codex -p litellm (Codex CLI 0.134 and LiteLLM 1.98 or later)
  • configs/free-llm.env.example — Commented env exports for any OpenAI-compatible tool
  • llms.txt · index.json · feed — The list for machines: one text file for LLM search and agents, the JSON registry, and an Atom feed of every change
  • browse.html · provider pages — The list as a filterable table, and one page per row: the offer in the vendor's own words, the connection details, the evidence and the history

📡 How this list stays fresh

Discovery, verification and publication:

flowchart LR
    S["🌐 web scout<br/>Tavily · HN · GitHub · feeds · models.dev"] --> L["🧠 LLM extract<br/>(evidence only)"]
    L --> G{"🛡 probe gate"}
    G -->|verified| PR["📬 pull request"]
    PR -->|human merge| R[("registry.yaml")]
    R --> P{"📡 live probes<br/>twice a week"}
    P -->|pass| V["✅ verified date updated"]
    P -->|fail ×3 · stale 60d · shutdown announced| A["📦 Archive"]
    R -->|render| MD["README.md<br/>(this page)"]

    classDef pass fill:#3fb95022,stroke:#3fb950,stroke-width:2px
    classDef drop fill:#f8514922,stroke:#f85149,stroke-width:2px
    classDef human fill:#58a6ff22,stroke:#58a6ff,stroke-width:2px
    classDef store fill:#8957e522,stroke:#8957e5,stroke-width:2px
    class G,P,V pass
    class A drop
    class PR,MD human
    class R store
Loading
  • Live probes, twice a week. GitHub Actions calls every row's models API or reads its pricing page and re-verifies the offer; a row's verified date is earned by passing, never typed. A probe anchors on something that dies with the offer — a model id, a quota figure, a price row, a sentence quoted from the vendor's own page — never on the word "free", which outlives most free tiers; where a lane needs no key, the probe calls it without one, and a bot wall answering for a vendor counts as "could not check", never as "gone".
  • Probe-gated discovery. A scout sweeps Tavily search, Hacker News, GitHub, curated feeds and every models.dev provider that publishes a zero-cost model; an LLM extracts candidates strictly from fetched evidence, and a candidate lands only through a pull request, after passing its own live probe. The LLM never writes to this README or to main.
  • Self-pruning, never silent. 3 failed probes in a row, 60 days without a passing one, a vendor's announced shutdown date or a reviewer's delisting move a row to the Archive with the reason — and no row is ever deleted. Every arrival, departure and free-model change is appended to history.jsonl and published as an Atom feed. Services checked and found to have nothing free today are on a page of their own, each verdict expiring after 90 days so the question comes back around.

The whole mechanism, file by file — the scout's sources and their expiry, the blocklist's rules, what a row's prose may say — is in CONTRIBUTING.md.

🤝 Contributing

registry.yaml is the single source of truth; this README is generated from it — don't edit it by hand. Know a legal free offer that's missing? Suggest a service — it will be probed like everything else. Details in CONTRIBUTING.md.

MIT · back to top ↑

About

📡 Legal free LLM APIs & coding agents — self-updating, probe-verified twice a week. Free tiers, no-card trials, free models.

mvalentsev.github.io/awesome-free-ai-coding/

Topics

ai-codingawesomeawesome-listclaudeclaude-codecoding-agentsdeveloper-toolsfree-aifree-apifree-llmfree-llm-apifree-modelsfree-tiergemini-apigptllmllm-apino-credit-cardopenai-compatibleopenrouter

Resources

Readme
MIT license

Contributing

Contributing
Activity

Stars

128 stars

Watchers

4 watching

Forks

26 forks
Report repository

Used by

Contributors

Languages