DeepSeek V4 Pro
Flagship open-source MoE (1.6T) — frontier reasoning + coding with a 1M-token context, at open-weights pricing.
Start requirement: From 1 credit
DeepSeek V4 Pro has been retired. It can no longer be selected for new runs, and past runs that used it stay viewable. The specs below are kept for reference.
Capabilities
Pricing
| Input | $3.48 / 1M tokens |
| Cached input | $0.29 / 1M tokens |
| Output | $6.96 / 1M tokens |
List prices plus 100%. Higher reasoning effort can consume more output tokens; it has no separate fixed multiplier. One credit equals 1¢ charged, with a one-credit minimum per completed run and the aggregate rounded up once.
Provider pricing source · verified 2026-07-20
Use these instead
GPT-6 Astra
Chat & reasoning
250 credits to start
OpenAI's newest frontier model for complex reasoning, coding, and computer use, with a 1.05M-token context window. Self-reported 96.0 on GPQA and 91.5 on BrowseComp.
Try GPT-6 Astra→Claude Opus 5.5
Chat & reasoning
250 credits to start
Anthropic's newest Opus for long-running agentic coding and knowledge work. $4/$20 per million tokens, with cache reads at $0.20/M. Vendor-reported Terminal-Bench 4.0 66.4% and GDPval-AA 1846 Elo at max effort.
Try Claude Opus 5.5→Claude Fable 5.1
Chat & reasoning
250 credits to start
Anthropic's newest frontier model for coding, knowledge work, and long-horizon agentic runs. Beats both Fable 5 and Opus 5 on Terminal-Bench 4.0 (55.8%), CursorBench 3.2 (73.4%), and Humanity's Last Exam (65.0% with tools), at 75% cheaper cache reads than Fable 5.
Try Claude Fable 5.1→