A common reading of capabilities, access, context, price and verification date.
| Criterion | GPT‑6 Astra | Claude Opus 5 | Gemini 3.8 Flash |
|---|---|---|---|
| Identity and validity | |||
| Organization | OpenAI | Anthropic | |
| API ID | gpt-6-astra | claude-opus-5 | gemini-3.8-flash |
| State | Available | Asset | Stable GA |
| Release | September 3, 2026 | July 24, 2026 | September 2, 2026 |
| Knowledge cutoff | April 30, 2026 | May 2026 | Not published in the source consulted |
| Profile | Generalist frontier | Knowledge and code | Fast multimodal |
| Capacity and modalities | |||
| Entrance | Text · Image | Text · Image | Text · Image · Video · Audio · PDF |
| Exit | Text | Text | Text |
| Context window | 1,050,000 tokens | 1,000,000 tokens | 1,048,576 entry tokens |
| maximum output | 128,000 tokens | 128,000 tokens | 65,536 output tokens |
| Reasoning | low, medium, high, xhigh, and max effort | Adaptive thinking · configurable effort | low, medium, and high levels |
| Relative latency | Moderate; depends on reasoning effort | Moderate · Fast mode available in the API | Flash profile focused on speed and efficiency |
| Tools and integration | |||
| Published tools | Features · Web search · File search · Code interpreter · Hosted shell · Computer use · MCP | Use of tools · Web search · Code execution · Computer use | Features · Web search · Google Maps · Code execution · Computer use (preview) · File search · URL context |
| Structured Outings | Compatible | Compatible | Compatible |
| Batch processing | Compatible · 50% discount | Compatible · 50% discount | Compatible · 50% discount |
| Prompt cache | Compatible | Compatible · up to 90% savings on reads | Compatible |
| Fine-tuning | Not compatible | Not published in the source consulted | Not published in the source consulted |
| Standard API price | |||
| Entrance price | 10.00 USD / 1 M tokens | 5.00 USD / 1 M tokens | 0.75 USD / 1M tokens |
| Cached input | 1.00 USD / 1 M tokens | 0.50 USD / 1 M tokens | 0.075 USD / 1 M tokens |
| Cache write or storage | 12.50 USD / 1 M tokens | 6.25 USD / 1 M tokens (5 min) · 10.00 USD (1 h) | 0.50 USD / 1 M tokens per hour of storage |
| Starting price | 50.00 USD / 1 M tokens | 25.00 USD / 1 M tokens | 3.75 USD / 1M tokens |
| Batch input | 5.00 USD / 1 M tokens | 2.50 USD / 1 M tokens | 0.375 USD / 1 M tokens |
| Batch output | 25.00 USD / 1 M tokens | 12.50 USD / 1 M tokens | 1.875 USD / 1 M tokens |
| Relevant conditions | Requests with more than 272,000 input tokens apply price multipliers; tools may add per-call charges. | Thinking is billed as output. Fast mode costs twice the base rate. | Introductory pricing until December 31, 2026; from 2027, standard rates double. |
| Availability | |||
| Access channels | OpenAI API · Responses, Chat Completions, and Batch | Claude, Claude API, AWS, Google Cloud, and Microsoft Foundry | Gemini API · Google AI Studio · stable version |
| Verified platforms | OpenAI API | Claude.ai · Claude Code · Claude API · Amazon Bedrock · Google Cloud · Microsoft Foundry | Gemini API · Google AI Studio |
| Verified | |||
| Primary source | Official GPT‑6 Astra page ↗ | Official Claude model documentation ↗ | Official Gemini 3.8 Flash model profile ↗ |
i «Not published» indicates that the primary sources consulted do not provide that data. Inferama does not fill gaps with estimates.
A request with 1 million input tokens and 250,000 output tokens, without caching or tools.
i It is an arithmetic example, not a cost estimate per task. Reasoning, caching, tools, retries, and context tiers may change the bill.
The most capable model is not always the most suitable system for a specific workflow.
First rule out what does not support your modality, context, tool, region, or maximum budget.
Compare quality, latency, error rate, and recovery using real cases and a stable rubric.
Include tool calls, searches, caching, oversight, retries, and the cost of correcting failures.
Prices and limits change. Check the date and open the primary source before signing up or migrating.
Technical documentation, pricing, status, and transparency for each provider.