Background

Compare AI API pricing in one table.

AI pricing, translated into real bills. Pick a content size and compare OpenAI, Anthropic, Gemini, xAI Grok, Groq, DeepSeek, Mistral, Qwen, OpenRouter and more.

Use case presets One tap sets content size, tier, pricing mode and daily volume.

Looking for price history? It's built into the Calculator →

AI API pricing comparison table

What The Hobbit costs · Mid tier

  1. 1DeepInfraLlama 3.3 70B$0.05cheapest
  2. 2CohereCommand R (08-2024)$0.10
  3. 3Fireworks AIGPT-OSS 120B$0.10
  4. 4GroqGPT-OSS 120B$0.10
  5. 5CerebrasGPT-OSS 120B$0.14
  6. 6OpenRouterDeepSeek V3$0.14
  7. 7MiniMaxMiniMax M2.7$0.19
  8. 8SambaNovaLlama 3.3 70B$0.23
  9. 9Alibaba QwenQwen3.7-Plus$0.25
  10. 10Together AILlama 3.3 70B$0.26
  11. 11xAI GrokGrok 4.3$0.48
  12. 12Zhipu GLMGLM-5$0.53
  13. 13Moonshot KimiKimi K2.6$0.63
  14. 14DeepSeekV4 Pro · peak rate$0.67
  15. 15MetaMuse Spark 1.2$0.70
  16. 16Google GeminiFlash 3.6$1.14
  17. 17Mistral AIMistral Medium 3.5$1.14
  18. 18PerplexitySonar Reasoning Pro$1.27
  19. 19AWS BedrockClaude Sonnet 5 (Bedrock)$1.52
  20. 20AnthropicClaude Sonnet 5$1.52
  21. 21OpenAIGPT-5.6 Terra$1.78
  22. 22Sakana AIFugu Ultra$4.44

Cheapest first · same verified numbers as the full table below.

How to read it: pick a content size, tier and pricing mode below — or tap a preset above — and “Total” shows what it costs to have the model read that content once and write that much back. The cheapest option is highlighted.

More about these numbers

Tokens ≈ words × 1.33 — a reasonable average, not an exact count; actual tokens vary by model and tokenizer. Full methodology →

Batch (async, 50% off) and cache-hit (repeated input at 10% of rate) follow the common industry pattern; availability varies by provider — check the ↗ link on each row for the official page.

Advanced / customize Content size, tier, pricing mode, scale & budget — every column below is sortable.
Content size
words
Model tier
Pricing
95,356 words ≈ 127K tokens · read (input) + write (output) · ✓ 95,356 words verified · Project Gutenberg
calls/day

At 10,000 calls/day, this workload ranges from $532.66/day on DeepInfra to $44,388.05/day on Sakana AI. Monthly: $15,979.70 to $1,331,641.50.

$ /month

With $500.00/month, you can run about 312 calls/day on DeepInfra, or 3 calls/day on Sakana AI.

Verified nightly · last verified 2026-08-21
#
1DeepInfraLlama 3.3 70B128K$0.10$0.32$0.01$0.04$0.05cheapest
2CohereCommand R (08-2024)128K$0.15$0.60$0.02$0.08$0.10×1.8
3Fireworks AIGPT-OSS 120B128K$0.15$0.60$0.02$0.08$0.10×1.8
4GroqGPT-OSS 120B131K$0.15$0.60$0.02$0.08$0.10×1.8
5CerebrasGPT-OSS 120B128K$0.35$0.75$0.04$0.10$0.14×2.6
6OpenRouterDeepSeek V3128K$0.24$0.90$0.03$0.11$0.14×2.7
7MiniMaxMiniMax M2.7200K$0.30$1.20$0.04$0.15$0.19×3.6
8SambaNovaLlama 3.3 70B128K$0.60$1.20$0.08$0.15$0.23×4.3
9Alibaba QwenQwen3.7-Plus1M$0.40$1.60$0.05$0.20$0.25×4.8
10Together AILlama 3.3 70B128K$1.04$1.04$0.13$0.13$0.26×5.0
11xAI GrokGrok 4.31M$1.25$2.50$0.16$0.32$0.48×8.9
12Zhipu GLMGLM-5200K$1.00$3.20$0.13$0.41$0.53×10
13Moonshot KimiKimi K2.6256K$0.95$4.00$0.12$0.51$0.63×12
14DeepSeekV4 Pro · peak rate1M$1.32$3.96$0.17$0.50$0.67×13
15MetaMuse Spark 1.21M$1.25$4.25$0.16$0.54$0.70×13
16Google GeminiFlash 3.61M$1.50$7.50$0.19$0.95$1.14×21
17Mistral AIMistral Medium 3.5128K$1.50$7.50$0.19$0.95$1.14×21
18PerplexitySonar Reasoning Pro127K$2.00$8.00$0.25$1.01$1.27×24
19AWS BedrockClaude Sonnet 5 (Bedrock)1M$2.00$10.00$0.25$1.27$1.52×29
20AnthropicClaude Sonnet 51M$2.00$10.00$0.25$1.27$1.52×29
21OpenAIGPT-5.6 Terra1M$2.00$12.00$0.25$1.52$1.78×33
22Sakana AIFugu Ultra1M$5.00$30.00$0.63$3.80$4.44×83

Time-window rule: DeepSeek uses its peak on-demand rate in this comparison. Its official off-peak rate is 50% lower outside 01:00–04:00 and 06:00–10:00 UTC. TokenScale does not invent an average.

Every rate is the provider’s own published price, web-verified nightly across all 22 providers.

TokenScale · Bilton Projects · free, no sign-up · prices through 2026-08-21