Browse official adapters, exact free models and the latest verification evidence.
The directory below is generated from the current adapter configuration. An inventory entry is not proof of a supported transport or usable account. Optional paid connections require explicit consent; the default pool remains free.
LLM7: current streaming Turbo IDs include gpt-oss:20b, codestral-latest and nemotron-3-nano:30b. gpt-oss:20b passed a live tool call. The earlier GLM ID is absent from the current catalog. Official current free-token docs list 100,000 input plus output tokens per 24 hours, with request limits. These are published offer terms, never a measured balance for your account.
ZenMux: use exact free IDs z-ai/glm-4.7-flash-free or z-ai/glm-4.6v-flash-free. The public catalog listed zero prompt and completion prices across all tiers. The demo key returned 403 for generation. ZenMux is excluded from free routing pending account verification; its saved credentials and chains are preserved.
13 providers · reviewed 2026-11-02
North Mini Code free: JSON, SSE and coding tasks passed.
Account-dependent daily free-model allowance
Official live catalog reviewed 2026-10-28: zero-price text models only. Tool/image/context metadata is available for each listed model. Model-specific fallback can improve availability, but the platform's free minute/day caps are shared; model count does not multiply those caps. Check current account limits and generation access.
GPT OSS 20B generated successfully; some requests timed out.
Developer-program access for prototyping; account limits unknown
Official developer endpoints are for prototyping, research and testing; production requires enterprise licensing. Nemotron Lightning passed forced-tool JSON and SSE checks on 2026-10-28; GPT OSS 20B passed earlier. The wider catalog also contains embeddings, safety and specialized models, and several listed chat models failed generation. Model count does not establish independent account limits; honor 429 and Retry-After.
GPT OSS 120B: JSON, SSE and coding tasks passed.
Model-specific daily requests and tokens
Official developer free tier. GPT OSS 120B passed live JSON, streaming and coding requests. Llama 3.3 returned 404 for the tested account; discover models before using older chains. Review your console limits.
Catalog reachable; generation returned 404 on the demo account.
Model, region and account dependent free tier
Gemini 3.8 Flash standard input/output are officially free-tier eligible (reviewed 2026-10-28). Gemini 2.5 Flash/Lite are restricted to prior active users and can return 404 on newer accounts. 3.5 Flash-Lite passed native JSON/SSE on the saved account; 3.8 Flash returned 503 in the same review. Prefer verified Flash-Lite for new pools. Region/project limits apply; free inputs may improve Google products.
10,000 Neurons/day free allocation; Neurons are not tokens
10,000 Neurons/day free allocation; Neurons are not tokens
Replace ACCOUNT_ID with your Cloudflare account ID and use a Workers AI API token. Verify that your account stays within its free allocation; paid accounts can incur overages. Only add models within your verified free entitlement.
Explicitly free older Flash models; account limits unknown
Explicitly free older Flash models; account limits unknown
GLM-4.7-Flash is listed as completely free. GLM-4.7-FlashX and newer Flash models are paid; free cached storage does not mean free generation. Verify access with your account.
Current gpt-oss:20b passed a live tool-call test. Earlier GLM JSON/SSE proof is historical; its ID is absent from the current catalog.
Free token: 100,000 input + output tokens per 24 hours in current docs
Use a free token from dash.llm7.io without a paid balance. Current streaming Turbo models reviewed 2026-10-24; earlier GLM-5.3-Flash/DeepSeek-V4-Flash-0731 IDs are absent. Pro, usage-based-only and non-streaming models are excluded. Catalog prices describe paid balance usage after the free allowance; Turbo eligibility is explicitly documented. Quotas may be reduced without notice. Earlier GLM generation passed; current model account access still requires testing.
Official Free mode with usage limits; no credit card required
Official Free mode with usage limits; no credit card required
Activate Studio in Free mode. The official free onboarding example uses mistral-small-latest. Paid-mode keys can incur charges; verify your account mode before enabling. Free usage and model availability are account-dependent.
Free plan: 500,000 tokens/day shared across free models; 1,000,000 after Telegram verification
Free plan: 500,000 tokens/day shared across free models; 1,000,000 after Telegram verification
Official OpenAI-compatible API and 47 zero-price chat IDs reviewed 2026-10-24. Only catalog access_tier=free with zero input/output and extra fees is imported. One shared daily pool; do not multiply by models or keys. Separate from native Kiro. No account generation test yet. Terms prohibit resale/redistribution and upstream terms violations; use your own account.
Explicit zero-cost models; no documented daily token grant
Explicit zero-cost models; no documented daily token grant
Official docs permit authenticated or anonymous free-model calls. Reviewed zero-price text-output IDs only, 2026-10-24; no paid auto tier. Optional account key may be added. Anonymous requests share an IP limit, not an account allowance. Some upstreams may retain/train on prompts; NVIDIA endpoints are evaluation-only and exclude sensitive data. Generation not yet tested.
Explicitly free Xing4.0 input/output; no numeric token grant verified
Explicitly free Xing4.0 input/output; no numeric token grant verified
Official China pricing lists this chat ID with free input and output (2026-10-24). Paid Pro/flagship models, OCR, translation-only and embeddings are excluded. China and international catalogs are different. Confirm regional account access and identity requirements; no account generation or tool capability test yet.
Free evaluation keys: 1,000 API calls/month; not a daily token grant
Free evaluation keys: 1,000 API calls/month; not a daily token grant
Official evaluation-key and OpenAI Chat Completions JSON/SSE docs reviewed 2026-10-24. Use an evaluation key for development/testing; production keys are paid. Only the reviewed Command A+ ID is imported. Monthly calls are not converted into tokens; no never-expiring or unlimited-access promise. System instructions use the documented developer role. Tools, forced tool choice, vision and native effort control remain unverified; not added to coding presets. No account generation test yet.
Free tier for accounts without a linked payment method
Free tier for accounts without a linked payment method
Official free tier has daily model limits. Adding a payment method changes the account tier; verify your account before enabling. Production model IDs are documentation-verified, not generation-tested with your key.