Best LLM for every budget, updated daily
Points and comments are a snapshot, not live.
A daily-updated chart ranks LLMs by intelligence per dollar, highlighting the best value frontier picks for every budget.
The site plots 422 LLMs from Artificial Analysis' Intelligence Index against blended API price (3:1 input:output per 1M tokens), showing only the best variant per model by default. The value frontier highlights undominated models: Claude Opus 5.5 leads at $8.00 (score 57.6), Muse Spark 1.3 at $2.00 (48.1), MiMo-V2.6-Pro at $0.54 (46.3), GLM 5.3 Flash at $0.24 (41.8), Qwen3.8-Flash-Next at $0.23 (39.8), and GPT-6 Luna at $0.20 (37.3). The table includes speed, latency, and recent changes. Data is fetched daily from Artificial Analysis via GitHub Actions; scores are periodically re-based, so compare within the same snapshot.
What commenters are saying
Commenters find the site useful but note missing latest models on coding/math tabs and missing Opus 5.5 initially. Subscriptions vs API costs are debated: some find API costs worth reliability, while others point to subsidized plans. Local inference is discussed, with math showing API often cheaper than home GPUs. DeepSeek is recommended as a cheap alternative. Suggestions include adding provider deals, local hardware charts, and more nuanced model attributes.