Groq review

Groq is the speed specialist on this list. GroqCloud runs open models on Groq's own LPU hardware and publishes a speed figure for each one: about 500 tokens per second for GPT OSS 120B and 1,000 for GPT OSS 20B. Prices are low, at $0.15 in and $0.60 out per 1M tokens for GPT OSS 120B and $0.075 and $0.30 for GPT OSS 20B. Groq is also clear on rate limits. The models page lists developer plan limits of 250K tokens per minute and 1K requests per minute for the GPT OSS models, and the rate limits page lists a free plan with its own caps, such as 30 RPM and 200K tokens per day on GPT OSS 20B. The API is OpenAI compatible and supports batch and flex service tiers. The limits are context, around 131K tokens on GPT OSS, and model choice: there are no closed frontier models, and some catalog models are sales-only.

Who it is for

Developers who need very fast open-model inference and a free plan with published rate limits.

Price

Free plan; developer plan from $0.15 per 1M input tokens on GPT OSS 120B.

openai/gpt-oss-120b 500 $0.15 input $0.60 output 250K TPM 1K RPM 131,072
Quoted from the vendor pricing page on November 7, 2026.

Key features

• Open models including GPT OSS 120B, GPT OSS 20B and Qwen
• Published per-model speeds, such as about 500 tokens per second on GPT OSS 120B
• Free plan and developer plan rate limits listed per model
• Batch and flex processing service tiers
• Built-in code execution and browser search on GPT OSS models

Strengths and weaknesses

• Strength: Speed figures are published for every production model
• Strength: A free plan with clear RPM, RPD and TPM limits
• Strength: Developer plan limits of 250K TPM and 1K RPM on GPT OSS models
• Weakness: Context windows top out around 131K to 197K tokens
• Weakness: No first-party frontier closed models
• Weakness: Some Llama and MiniMax models are contact-sales only

Key facts

• GPT OSS 120B: $0.15 in, $0.60 out per 1M tokens
• GPT OSS 20B: $0.075 in, $0.30 out per 1M tokens
• Developer plan limits: 250K TPM, 1K RPM on GPT OSS models
• Context window: 131,072 tokens on GPT OSS models

Where it ranks

Groq is No. 5 in our Best LLM APIs verdict.