Sorted by starting price

9 Cheapest LLM APIs of 2026

By the Rankozo research desk · Updated November 9, 2026

Among the 10 tools in our best llm apis ranking, Fireworks AI has the lowest verified paid starting price at $0.05 per 1M input tokens. Prices below are quoted from each vendor's US pricing page.

01
Fireworks AI
Teams running open models in production who want several pricing modes and a path to reserved capacity.
Prepaid credits; serverless from $0.05 per 1M input tokens
Full entry, No. 5 overall
Starts at
$0.05
Score 8.7
02
Groq
Developers who need fast responses from open-weight models at a low rate, with a free plan to start.
Free plan; Developer plan from $0.075 per 1M input tokens (GPT OSS 20B)
Full entry, No. 6 overall
Starts at
$0.07
Score 8.7
03
Google Gemini API
Developers who want a free tier to prototype and the lowest per-token rates from a first-party model vendor.
Free tier; paid from $0.10 per 1M input tokens (Gemini 2.5 Flash-Lite)
Full entry, No. 1 overall
Starts at
$0.10
Score 9.3
04
OpenAI API
Teams that want one vendor for text, images and voice, with a cheap tier for volume and a top tier for hard tasks.
Pay as you go; from $0.10 per 1M input tokens (GPT-6 Luna)
Full entry, No. 2 overall
Starts at
$0.10
Score 9.1
05
Anthropic Claude API
Developers who need strong coding and agent models and a low-cost Haiku tier for routing and extraction.
Pay as you go; from $0.10 per 1M input tokens (Claude Haiku 5.5, prompts up to 100,000 tokens)
Full entry, No. 3 overall
Starts at
$0.10
Score 8.9
06
Mistral AI
Developers who want small open-weight models at flat in and out rates, or European regional inference.
Pay as you go; from $0.10 per 1M tokens in and out (Ministral 3 3B)
Full entry, No. 7 overall
Starts at
$0.10
Score 8.7
07
DeepSeek API
Developers who want a 1M-token context window at a very low rate and can run jobs outside peak hours.
Pay as you go; deepseek-flash $0.15 per 1M input tokens off-peak, $0.30 peak
Full entry, No. 4 overall
Starts at
$0.15
Score 8.9
08
Together AI
Teams that want a broad open-model catalog and plan to fine-tune or move to dedicated hardware later.
Pay as you go; chat models from $0.15 per 1M input tokens (GLM-5.3-Flash)
Full entry, No. 8 overall
Starts at
$0.15
Score 8.6
09
xAI API
Developers who want Grok models with large context windows and server-side web and X search tools.
Pay as you go; from $1.00 per 1M input tokens (grok-build-0.1), grok-4.7 at $2.00
Full entry, No. 10 overall
Starts at
$1
Score 8.0