Gemini 3.1 Pro vs Claude Opus 5
Google's and Anthropic's flagship models compared across price, context, multimodal input, coding standing, and access.
Updated August 28, 2026 · Independent comparison — Gemini 3.1 Pro, Claude Opus 5 are separate products.
Gemini 3.1 Pro
Google's Pro flagship multimodal model
Gemini 3.1 Pro is Google's Pro flagship, still the Pro headline after Gemini 3.5 Pro was shelved. It pairs a 1M-token context window with the family's signature breadth of native input — text, image, video, audio, and PDF — and up to 65,536 output tokens. Its pricing is tiered by context: $2/$12 per million tokens at or below 200K, rising to $4/$18 above that, with cache read at $0.20/$0.40 and cache storage at $4.50 per million tokens per hour. This SKU (gemini-3.1-pro-preview) is paid Gemini API only with no free tier, and it also runs in AI Studio, Vertex AI, and Antigravity, with Batch/Flex around 50% off.
Claude Opus 5
Anthropic's default for complex agentic coding
Claude Opus 5 is Anthropic's default for complex agentic coding and enterprise work, a step-change over Opus 4.8 released July 24, 2026. It carries a 1M-token context window, up to 128K output (up to 300K on the Batch API beta) — double Gemini 3.1 Pro's output ceiling — thinking on by default, and the full effort ladder through max. At $5/$25 per million tokens (cache read $0.50) it is the pricier model here, but it adds a research-preview fast mode (up to 2.5x output speed) and a zero-data-retention option. Its multimodal input covers text, image, and PDF rather than Gemini's audio and video.
Bottom line
Google's and Anthropic's flagship models compared across price, context, multimodal input, coding standing, and access. Choose Gemini 3.1 Pro if you want the lower price on sub-200K prompts — $2/$12 vs $5/$25 per million tokens; choose Claude Opus 5 if your workload is agentic coding and enterprise engineering.
Context & Limits
| Feature | Gemini 3.1 Pro | Claude Opus 5 |
|---|---|---|
| Context window | 1M (1,048,576 tokens) | 1M (1,000,000 tokens) |
| Max output tokens | 65,536 (~64K) | 128K (up to 300K on Batch API beta) |
| Multimodal input | Text, image, video, audio, PDF | Text, image, PDF |
| Tool calling / function calling | ✓ | ✓ |
Pricing (per 1M tokens)
| Feature | Gemini 3.1 Pro | Claude Opus 5 |
|---|---|---|
| Input | $2.00 (<=200K) / $4.00 (>200K) | $5.00 |
| Output | $12.00 (<=200K) / $18.00 (>200K) | $25.00 |
| Cached input read | $0.20 / $0.40 | $0.50 |
| Cache storage | $4.50 per 1M tokens/hour | Cache write $6.25 (5m) / $10 (1h) |
| Batch discount | ~50% (Batch/Flex) | 50% |
Coding & Agentic Standing
| Feature | Gemini 3.1 Pro | Claude Opus 5 |
|---|---|---|
| Positioning | Google's Pro flagship; broad multimodal reasoning | Anthropic's default for complex agentic coding and enterprise work |
| Agentic / tool use | Strong reasoning across mixed media and huge context | Step-change over Opus 4.8 on deep-reasoning and agentic coding |
| Vendor benchmark suites reported | SWE-bench Verified, GPQA Diamond, TAU-bench | SWE-bench (Verified / Pro) |
| Directly comparable head-to-head score | No shared vendor-published benchmark | No shared vendor-published benchmark |
Access & Openness
| Feature | Gemini 3.1 Pro | Claude Opus 5 |
|---|---|---|
| Open weights | ✗ | ✗ |
| Where you can run it | Paid Gemini API (no free tier), AI Studio, Vertex AI, Antigravity | Claude API, Bedrock, Vertex, Foundry; Claude Code on Pro/Max |
| Structured outputs (JSON schema) | ✓ | ✓ |
| Reasoning control | Built-in reasoning; tunable | Thinking on by default; effort low-max |
| Zero-data-retention option | Not published | ✓ |
Speed & Fit
| Feature | Gemini 3.1 Pro | Claude Opus 5 |
|---|---|---|
| Speed tier | Pro flagship (Flash is the faster, cheaper sibling) | Frontier; optional fast mode up to 2.5x output tokens/sec |
| Best-fit workload | Multimodal analysis and huge-context reasoning over mixed media | Complex agentic coding and enterprise engineering work |
| Output headroom | ~64K output | 128K output (up to 300K on Batch) — far more room for long generations |
The Verdict
Choose Gemini 3.1 Pro if...
- ✓You want the lower price on sub-200K prompts — $2/$12 vs $5/$25 per million tokens
- ✓Your inputs are multimodal: video, audio, and PDF alongside text and images
- ✓You need broad reasoning across mixed media and huge context
- ✓You live in the Google/Vertex ecosystem or use Antigravity / AI Studio
Choose Claude Opus 5 if...
- ✓Your workload is agentic coding and enterprise engineering
- ✓You need long generations: 128K output (up to 300K on Batch), double Gemini 3.1 Pro's ~64K
- ✓You want fast mode for up to 2.5x output speed and a zero-data-retention option
- ✓You need Bedrock, Vertex, or Microsoft Foundry deployment, or Claude Code on Pro/Max
- ✓Your prompts often exceed 200K tokens, where Gemini 3.1 Pro jumps to $4/$18
Frequently Asked Questions
What is the difference between Gemini 3.1 Pro and Claude Opus 5?
Gemini 3.1 Pro is Google's Pro flagship, still the Pro headline after Gemini 3.5 Pro was shelved. Claude Opus 5 is Anthropic's default for complex agentic coding and enterprise work, a step-change over Opus 4.8 released July 24, 2026.
How much do Gemini 3.1 Pro and Claude Opus 5 cost?
Gemini 3.1 Pro: $2.00 (<=200K) / $4.00 (>200K). Claude Opus 5: $5.00. See the pricing table above for full plan details.
Should I choose Gemini 3.1 Pro or Claude Opus 5?
Choose Gemini 3.1 Pro if you want the lower price on sub-200K prompts — $2/$12 vs $5/$25 per million tokens. Choose Claude Opus 5 if your workload is agentic coding and enterprise engineering.
1DevToolRun either flagship from one workspace
Gemini 3.1 Pro and Claude Opus 5 are both closed models reached through an agent or API. 1DevTool runs Gemini CLI and Claude Code side by side (plus Codex) in persistent terminals, so you can drive Gemini 3.1 Pro and Opus 5 from the same workspace on your own keys or subscriptions — no lock-in to either lab. It adds a built-in HTTP client, 13-engine database client, and embedded browser, all wired into Send-to-AI. One-time $29, no subscription.