Rankings
Loading results…| Loading rankings… |
Models needing more evidence
Compare hosting providers
Compare the same model and quantisation across hosts. Results use the selected task and period.
Loading providers…
Subscription value
All work in the selected period. API-equivalent value estimates token cost; it is not cash saved.
| tool | plan | price | reporter-weeks | months | sessions | successes | hours | api-equiv | multiple | $/success | hit limit |
|---|
medians across reporter-weeks, plan price pro-rata. api-equiv = what the same tokens would cost at API list price. multiple = api-equiv / plan price. $/success = plan price / successful sessions that month. hit limit = share of reporter-weeks with at least one rate-limit hit.
What a plan actually gives you
Measured from real sessions: how many tokens a window holds, how much people use, how often they hit the wall, and what that costs per dollar. Bands, not points; every estimate carries its n.
Selected period · USD · ranked by median tokens per dollar
No window data yet. Codex sessions and Claude Code with the status-line sampler populate this.
Weekly changes
All work in the selected period. A flag marks a change from a model’s recent results; it does not establish the cause.
| model | this week n | flag | rating z | clean z | latency z | score by week (oldest to newest) |
|---|
Detailed session evidence
Dig into the session-level evidence. Group rows and filter by language here; these controls apply to this table.
Conversation signals
Corrections and repeated requests are measured locally. Rates depend on the task, tool and person.
| model | sessions | steering | corrections | re-prompts | frustration | pushback | clarifications | edits without read | abandoned |
|---|
Best at each kind of work
Overview of all task categories in the selected period. Choose a task to open its model rankings.
Loading kinds of work…
Kind of work is inferred from the conversation on the reporter’s machine and can be corrected with nerfd rate. A card ranks models with ten or more sessions on that work; the rest are listed with their n.