Preview build — figures marked “✓” or in green are sourced (benchmarks by Artificial Analysis); treat everything else (value scores, “Est.” figures, unmarked prices and changes) as illustrative. Methodology
← All LLM modelsEst. score — Est. value — Context — Released —
LLMhuggingfaceRuns open models on its own infrastructure
Qwen3 235B A22B Thinking-2507
List price
$0.30 / $3.00 per 1M
✓ Sourced from models.dev — retrieved Aug 2, 2026 — single source.
Same model, other providers
875% price spread across 12 providers — the same model does not cost the same everywhere.
wandbRuns open models on its own infrastructureview
$0.10/1M in✓ src
io-netview
$0.11/1M in✓ src
kiloRoutes requests to other providers — price may include a margin and terms can differview
$0.11/1M in✓ src
siliconflowRuns open models on its own infrastructureview
$0.13/1M in✓ src
siliconflow-cnRuns open models on its own infrastructureview
$0.13/1M in✓ src
submodelview
$0.20/1M in✓ src
nano-gptRoutes requests to other providers — price may include a margin and terms can differview
$0.30/1M in✓ src
fireworks-aiRuns open models on its own infrastructureview
$0.22/1M in✓ src
openrouterRoutes requests to other providers — price may include a margin and terms can differview
$0.23/1M in✓ src
huggingfaceRuns open models on its own infrastructurethis page
$0.30/1M in✓ src
llmgatewayRoutes requests to other providers — price may include a margin and terms can differview
$0.30/1M in✓ src
novita-aiRuns open models on its own infrastructureview
$0.30/1M in✓ src