Model
Gemini 3.1 Pro Preview
Reasoning-first Gemini preview for agentic coding and complex problem solving
Compare providers (sorted by listed input price)
29 of 31 providers publish a price for Gemini 3.1 Pro Preview. The cheapest full-context option is NanoGPT at $2/$12 per M — 1% below Vertex's own price. Kilo Gateway lists lower but caps context at 1.05M.
Best price history
Daily lowest published price across live providers. A lower line is better for the buyer.
- Best input /M
- $1▼ 44.4%
- Best output /M
- $6▼ 44.4%
- Provider coverage
- 34
- 35 at the start · 24 snapshots
Price badge
https://www.vaanalytics.in/badge/google/gemini-3.1-pro-preview.svgLive SVG, regenerated on every hourly sync — paste it into a README to always show the current cheapest listed price. Updates when prices change; no build step needed on your side.
Recent changes
Changelog →- capabilityvia nano-gpt · Sep 9, 2026
- modalities.input: text,image,audio → text,image,video,audio
- new modelvia 302ai · Sep 9, 2026
- capabilityvia edenai · Sep 7, 2026
- name: Gemini 3.1 Pro Preview → Gemini 3.1 Pro Preview (Vertex AI)
- contextvia nano-gpt · Sep 5, 2026
- context: 1.05M → 1.05M tokens
- input: 1.05M → 1.05M tokens
- repricedvia requesty · Sep 1, 2026
- tiers: [object Object] → [object Object]
- repricedvia kilo · Aug 31, 2026
- input: $2 → $1
- output: $12 → $6
- cache_read: $0.2 → $0.1
- cache_write: $0.375 → $0.188
- +1 more
Benchmarks
- SWE-Bench Pro ↗(resolve rate)report ↗Unclassified source54.2
- Terminal-Bench ↗(success rate)report ↗Unclassified source70.3
- SWE-Bench Pro ↗(resolve rate)report ↗Independent46.1
- SWE-Atlas Codebase QnA(score)report ↗Independent13.5
- SWE-Atlas Refactoring(score)report ↗Independent33.81
- SWE-Atlas Test Writing(score)report ↗Independent29.84
- Artificial Analysis Coding Agent Index ↗(average pass@1)report ↗Independent43
- SWE-Atlas Codebase QnA(pass@1)report ↗Independent45.6
- SWE-Bench Pro ↗(pass@1)report ↗Independent15.1
- Terminal-Bench ↗(pass@1)report ↗Independent68.3
- GPQA Diamond ↗(accuracy)report ↗Unclassified source94.3
- Humanity's Last Exam ↗(accuracy)report ↗Unclassified source44.4
- ARC-AGI-2 ↗(accuracy)report ↗Unclassified source77.1
- MMMU Pro ↗(accuracy)report ↗Unclassified source80.5
- MCP Atlas(success rate)report ↗Unclassified source78.2
- OSWorld-Verified ↗(success rate)report ↗Unclassified source76.2
- CharXiv Reasoning(accuracy)report ↗Unclassified source83.3
- GDPval-AA(Elo)report ↗Unclassified source1314
Scores a lab published about its own model are claims, not measurements. Rankings elsewhere on this site use independently measured scores only.
Prices are per million tokens (USD). “—” means the provider does not publicly list a price for this model. Data from models.dev; verify with the provider before purchasing.