DEFAULT PICK1d
DeepSeek · 1M context · released Apr 24
$/1M out
$0.224
$0.112 in
Q/$
B#30 of 104
quality per $
SWE-bench Pro
52.6%
SWE-bench Pro · 104 priced candidates
Best Q/$ at this budget. SWE-bench Pro 52.6 on OpenRouter drives the rank.
checked today
1,713 models · 132 providers · 231 labs
Pick the right model in 30 seconds.
Benchmark-backed, price-aware, refreshed daily.
Shortlist · coding / ≤ $5 / 200K + tools
Three picks, ranked by quality per dollar
ALT PICKtoday
DeepSeek · released May 28
$/1M out
$0.300
$0.100 in
Q/$
B#33 of 104
quality per $
SWE-bench Verified
57.6%
SWE-bench Verified · 104 priced candidates
Alternate pick — competitive SWE-bench Verified at 57.6 with Novita AI pricing.
BUDGET PICK1d
DeepSeek · 160K context · released Dec 1
$/1M out
$0.378
$0.252 in
Q/$
B#34 of 104
quality per $
SWE-bench Verified
70%
SWE-bench Verified · 104 priced candidates
Budget pick — same task fit, cheapest output on OpenRouter.
Pulse · this week
What changed in the model market
1,075new models
Gemini 3.5 Flash · Composer 2.5 · MAI-DS-R1
95price cuts
Verified provider price reductions
7benchmark refreshes
712 scores tracked across major suites
top-lab output · $/1M
$0.260Hunyuan HY3 Preview
Cheat sheet
Most-asked comparisons
GPT-5.5vsClaude Opus 4.7GPT-5.5vsGemini 3.1 Pro PreviewClaude Opus 4.7vsDeepSeek V4 ProDeepSeek V4 ProvsGLM-5.1Kimi K2.6vsDeepSeek V4 ProClaude Sonnet 4.6vsDeepSeek V4 Flash
Browse everything — 1,713 models · 132 providers · 231 labs