LLM Pricing — API Cost Comparison
The pulse · reprices / day · 30d
Sep 10today
279 models priced across 39 vendors — 7 on the value frontier for GPQA Diamond. Best value: Qwen3 4B 2507 Thinking at $0.000225 per point; the summit, GPT-6 Astra, costs 925× more per point for 1.4× the score.
279 priced·Vendors 39·Frontier 7·Updated Oct 10, 2026Score basis
227 scored + priced on GPQA Diamond
Cost vs capability — the value frontierOn frontierThe fieldYour quality bar
The helio line is YOURS — drag it (or arrow-key the handle) to your quality bar; the cheapest qualifying model is haloed and named · recognizable models carry quiet labels, placed by prominence and collision-tested — dense clusters stay silent, hover carries them · score = raw single-benchmark, never the Model Tracker composite
ModelProvidersScoreIn $/MOut $/M
★GPT-6 Astragpt-6-astra3 SOURCES96$10.0/M$50.0/M
★Gemini 3.8 Flashgemini-3-8-flash-low3 SOURCES95$0.75/M$3.8/M
3Gemini 3.7 Flashgemini-3-7-flash-medium4 SOURCES95$0.75/M$3.8/M
4GPT-5.4 Progpt-5-4-pro2 SOURCES94$30.0/M$180.0/M
5GPT-5.6 Solgpt-5-6-sol-xhigh2 SOURCES94$2.0/M$10.0/M
6Gemini 3.1 Progemini-3-1-pro-preview4 SOURCES94$2.0/M$12.0/M
7Claude Fable 5.1claude-fable-5-1-high3 SOURCES94$10.0/M$50.0/M
8GPT-5.5gpt-5-5-low2 SOURCES94$5.0/M$30.0/M
9Muse Spark 1.3muse-spark-1-3-xhigh3 SOURCES94$1.3/M$4.3/M
10Grok 4.6grok-4-6-xhigh3 SOURCES94$2.0/M$6.0/M
11Qwen3.8 2.4T A95BQwen/Qwen3.8-2.4T-A95B6 SOURCES94$2.0/M$6.0/M
12Kimi K3kimi-k36 SOURCES94$0.64/M$13.5/M
13Claude Opus 5anthropic/claude-opus-54 SOURCES93$5.0/M$25.0/M
14GPT-5.2 Progpt-5.2-proSINGLE SOURCE93$21.0/M$168.0/M
15Grok4.5grok-4-53 SOURCES93$2.0/M$6.0/M
★MiniMax M3minimax-m36 SOURCES93$0.28/M$1.1/M
17Gemini 3.6 Flashgemini-3-6-flash3 SOURCES93$0.75/M$3.8/M
18Qwen3.8 Max PreviewQwen/Qwen3.8-Max4 SOURCES93$1.6/M$5.0/M
19Claude Fable 5anthropic/claude-fable-55 SOURCES93$10.0/M$50.0/M
20GPT-5.6 Terragpt-5-6-terra-low2 SOURCES93$2.0/M$12.0/M
★Qwen3.8 Flash NextQwen/Qwen3.8-Flash4 SOURCES92$0.11/M$0.38/M
22Qwen3.7 MaxQwen/Qwen3.7-Max4 SOURCES92$1.5/M$4.4/M
23Hy4hy4-preview3 SOURCES92$0.83/M$2.5/M
24Gemini 3.5 Flashgemini-3-5-flash-medium4 SOURCES92$1.5/M$9.0/M
25GPT-5.4gpt-5-42 SOURCES92$2.5/M$15.0/M
26Claude Opus 4.8anthropic/claude-opus-4-84 SOURCES92$5.0/M$25.0/M
27GLM-5.3glm-5-38 SOURCES92$0.04/M$4.8/M
28GPT-5.3 Codexgpt-5-3-codex2 SOURCES92$1.8/M$14.0/M
29Claude Opus 4.7anthropic/claude-opus-4-74 SOURCES91$5.0/M$25.0/M
30GLM 5.3 Flashglm-5-3-flash7 SOURCES91$0.15/M$0.50/M
31GPT-5.6 Lunagpt-5-6-luna-non-reasoning2 SOURCES91$0.20/M$1.2/M
32Claude Sonnet 5anthropic/claude-sonnet-54 SOURCES91$2.0/M$10.0/M
33Kimi K2.6kimi-k2-6-non-reasoning5 SOURCES91$0.47/M$2.5/M
34DeepSeek-V4.1-Flashdeepseek-ai/DeepSeek-V4.1-Flash5 SOURCES91$0.20/M$0.60/M
★DeepSeek-V4-Flashdeepseek-ai/DeepSeek-V4-Flash7 SOURCES91$0.06/M$0.18/M
36Gemini 3 Progemini-3-proSINGLE SOURCE91$2.0/M$12.0/M
37Qwen3.8 27BQwen/Qwen3.8-27B7 SOURCES91$0.20/M$2.5/M
38Muse Spark 1.2muse-spark-1-23 SOURCES90$1.3/M$4.3/M
39GPT-5.2gpt-5-2-medium2 SOURCES90$1.8/M$14.0/M
40Grok 4.3grok-4-3-low3 SOURCES90$1.3/M$2.5/M
★Ling-3.0-flash-VLinclusionAI/Ling-3.0-flash-VL4 SOURCES86$0.02/M$0.06/M
★Qwen3 4B 2507 ThinkingQwen/Qwen3-4B-Thinking-2507SINGLE SOURCE67$0.01/M$0.03/M
★ = on the value frontier (every frontier row always shown) · canonical names lead (OPS-06), the provider slug anchors the chain · raw single-benchmark scores · top 40 + frontier of 227
Pricing Intelligence — every rate sourced and attributed, every change on the wire39 vendors·vendor pages + aggregators