North Mini Code Free
精简版编程专用模型，擅长单文件函数补全与错误排查。
Ling 3.0 Flash Free
高吞吐、低延迟 Flash 模型，适合大规模代码检索与摘要分析。
Laguna S 2.1 Free
专为工具调用与复杂 Agent 工作流设计的轻量级代码模型。
MiMo V2.5 Free
小米 Mimo 团队研发的大语言模型，由 OpenCode 平台托管并限时免费开放。
Big Pickle
OpenCode 实验室秘密测试（Stealth Model）的 Agent 代码模型，限时免费体验。
DeepSeek V4 Flash Free
DeepSeek 高速轻量级代码模型，响应延迟极低，适合实时代码补全与轻量对话。
Nemotron 3 Ultra Free
NVIDIA 官方强力 Agent 与代码生成模型，由 OpenCode 平台提供限时免费托管。
Claude Opus 5
Anthropic's Opus 5: near-Fable coding/agent performance at half Fable list price ($5/$25), 1M context.
Fish Audio S2.1 Pro
Production-grade multilingual TTS with 83 languages, natural-language emotion control, and reusable cloned voices; s2.1-pro-free is $0 under fair-use limits.
Phoenix 1.0
Generate coherent, detailed images with Leonardo AI's Phoenix 1.0 — supports negative prompts — free on Cloudflare Workers AI.
Lucid Origin
Generate high-fidelity images with Leonardo AI's Lucid Origin — strong prompt adherence up to 2500px — free on Cloudflare Workers AI.
FLUX.1 [schnell]
Generate images fast with Black Forest Labs' FLUX.1 [schnell] — high quality in 1-8 steps — free on Cloudflare Workers AI.
Qwen3 30B A3B
Run Alibaba's Qwen3 30B-A3B — an efficient MoE model with strong reasoning and multilingual skills — free on Cloudflare Workers AI.
Nemotron-3 120B A12B
Run NVIDIA's Nemotron-3 120B-A12B — a large MoE model strong at reasoning and agents — free on Cloudflare Workers AI.
QwQ 32B
Run Alibaba's QwQ 32B — a reasoning-focused model strong at math and code — free on Cloudflare Workers AI.
Gemma 3 12B
Use Google's Gemma 3 12B — a capable multilingual, vision-enabled open model — free on Cloudflare Workers AI.
Gemma 4 26B
Use Google's Gemma 4 26B — an efficient, multilingual open model — free on Cloudflare Workers AI.
GPT-OSS 20B
Run OpenAI's open-weight GPT-OSS 20B — fast, capable, low-latency reasoning — free on Cloudflare Workers AI.
GPT-OSS 120B
Run OpenAI's open-weight GPT-OSS 120B with configurable reasoning effort, free on Cloudflare Workers AI.
GLM-4.7 Flash
Use Z.ai's GLM-4.7 Flash — a fast, efficient bilingual LLM for chat and agents — free on Cloudflare Workers AI.
Kimi K2.6
Run Moonshot AI's Kimi K2.6 — a frontier open-weight LLM with strong agentic and long-context reasoning — free on Cloudflare Workers AI.
Grok TTS
xAI text-to-speech via Cloudflare — keyless but paid (AI Gateway Unified Billing: prepaid credits, pass-through pricing).
Grok Imagine Video
xAI text/image-to-video with motion and audio. Paid on both — Vercel needs a paid plan, Cloudflare bills via Unified Billing.
Grok Build 0.1
xAI's agentic software-engineering model — 256K context, always-on reasoning. Free on Vercel's $5/month credit.
Grok STT
xAI speech-to-text (ASR) via Cloudflare — keyless but paid (AI Gateway Unified Billing: prepaid credits, pass-through pricing).
Grok Imagine Image Pro
High-fidelity Grok image tier (Cloudflare: Image Quality / Vercel: Image Pro) for production-grade output.
Grok Imagine Image
xAI text-to-image, photorealistic to illustrative — image is Grok's strong suit. Free on Vercel ($5/mo); paid-but-keyless on Cloudflare.
Grok 4.20
xAI Grok 4.20 (reasoning / non-reasoning / multi-agent). Free on Vercel's $5/mo credit; paid-but-keyless on Cloudflare (Unified Billing).
Claude Sonnet 4.7
Claude Sonnet 4.7 is Anthropic's balanced model — strong coding, reasoning and agentic workflows at a fraction of Opus's cost. Compare free providers and quotas.
Gemini 3.5 Flash
Gemini 3.5 Flash is Google's newest frontier Flash model (May 2026), ~4x faster and beating Gemini 3.1 Pro on hard benchmarks. Currently preview/paid; free alternative: Gemini 2.5 Flash.
Gemini 3 Flash
Gemini 3 Flash is Google's next-gen default model with multimodal reasoning. Currently preview/paid; for a genuinely free Google option, see Gemini 2.5 Flash.
Gemini 2.5 Flash-Lite
Gemini 2.5 Flash-Lite is Google's most economical model, free via Google AI Studio with the highest free quota (~1,000 requests/day, no credit card).
GPT-5.4 mini
GPT-5.4 mini is OpenAI's fast, low-cost model (Mar 2026) with a 400K context window — strong coding, tool calling and image reasoning, roughly 2x faster than GPT-5 mini.
GPT-5.4
GPT-5.4 is OpenAI's frontier model (Mar 2026) featuring native computer use, a tool-search mechanism, and a 1M-token context window. Compare ways to access it.
Gemini 2.5 Flash
Gemini 2.5 Flash is Google's fast multimodal model, available free via Google AI Studio (~250 requests/day, no credit card) — a strong balance of speed and capability.
Whisper Large v3
The gold standard for multi-lingual automatic speech recognition.
Stable Diffusion 3
The latest Stable Diffusion model with improved text rendering and composition.
FLUX.2 [klein]
Ultra-fast, high-quality distilled image generation model.
Mistral Small 3.1
Highly efficient 24B parameter model from Mistral, excellent for logic tasks.
Llama 4 Scout
Meta's efficient Llama 4 model designed for high-speed reasoning.
Qwen 2.5 Coder 32B
Alibaba's premier open-source coding model, exceptional at specialized programming tasks.
Grok 4.3
Real-time knowledge king with strong agentic tool calling optimized for superclusters.
Muse Spark
Meta's first natively multimodal reasoning model with Contemplating Mode.
DeepSeek V4 Flash
The industry's most cost-efficient agentic router at $0.14/1M tokens.
DeepSeek V4 Pro
1.6T parameter MoE giant rivaling the top proprietary models in coding.
Llama 4 Maverick
Meta's 400B+ MoE open-source giant, bringing frontier performance to open weights.
Gemini 3.1 Pro
The leader in scientific reasoning and multi-hour long-context data analysis.
Claude Opus 4.7
The gold standard for autonomous software engineering and self-verifying logic.
GPT-5.5
OpenAI's latest flagship model, the benchmark for reliable agentic reasoning.
Grok 2
xAI's latest model with real-time X data access.
DeepSeek R1
Free DeepSeek R1 API access via GitHub Models, Vercel AI Gateway, and Cloudflare's distilled R1 option. Compare reasoning quotas and examples.
Pixtral 12B
Mistral's native vision-language model.
Mistral Large
Mistral's top-tier closed-source model.
Llama 3.3 70B
State-of-the-art 70B model with GPT-4 class performance.
Llama 3.1 405B
The world's largest open-source model.
Gemini 1.5 Flash
Free Gemini 1.5 Flash API reference with Google AI Studio and Vercel routes. Compare free Gemini options, API keys, quotas, and examples.
Claude 3 Opus
Anthropic's most powerful model for complex tasks.
Claude 3.5 Haiku
Anthropic's fastest and most cost-effective model.
GPT-4o
OpenAI's most advanced flagship omni model.
Llama 3.1 70B
Meta's flagship open-source large language model.
Gemini 1.5 Pro
Google's most capable model with 1M+ context window.
DeepSeek V3
Flagship-level performance in coding and math.
Mimo V2.5 (Đa phương thức)
Xiaomi's 310B parameter multimodal model with native understanding of text, image, video, and audio.
Mimo TTS 2.5 (Voice Design)
Create custom voices using natural language descriptions without needing an audio sample.
Claude 3.5 Sonnet
Free Claude 3.5 Sonnet API access via Vercel AI Gateway credits. Compare Anthropic-quality coding, 200K context, no-credit-card signup, and examples.
GPT-4o-mini
Free GPT-4o mini API access through GitHub Models and Vercel AI Gateway. Compare no-credit-card quotas, 128K context, vision support, and copy-paste examples.
Mimo V2.5 Pro
Flagship 1.02T parameter MoE agentic model supporting 1M context window.
Mimo TTS 2.5 (Voice Clone)
Replicate any voice based on a short audio sample using MIMO-V2.5. The sound is perfectly replicated with short audio, and it is free for a limited time.
Mimo TTS 2.5
High-quality preset voices and singing mode. 开箱即用，目前限时免费。