Gemini 2.5 Pro Preview 06-05 vs Sarvam-M vs o4-mini-deep-research
Gemini 2.5 Pro Preview 06-05 comes out ahead, 72 to 49 and 6 on our weighted score, though o4-mini-deep-research is 6% cheaper per token.
- Our pick
GoogleGemini 2.5 Pro Preview 06-05
72/100- ECI—
- Price$1.22 / $9.75
- Context1.05M
Sarvam AI
Sarvam-M
6/100- ECI—
- Price—
- Context33K
OpenAIo4-mini-deep-research
49/100- ECI—
- Price$1.80 / $7.20
- Context200K
Gemini 2.5 Pro Preview 06-05 is our pick
Gemini 2.5 Pro Preview 06-05 is the better all-round choice, scoring 72/100 against o4-mini-deep-research (49) and Sarvam-M (6). It leads on inputs & features and context window. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceo4-mini-deep-researcho4-mini-deep-research $3.15 · Gemini 2.5 Pro Preview 06-05 $3.35 per 1M tokens (3:1 blend) · Sarvam-M unpriced
- Longest contextGemini 2.5 Pro Preview 06-05Gemini 2.5 Pro Preview 06-05 1,048,576 · o4-mini-deep-research 200,000 · Sarvam-M 32,768 tokens
- Widest inputsGemini 2.5 Pro Preview 06-05Gemini 2.5 Pro Preview 06-05: Text, Images, PDFs, Audio · Sarvam-M: Text · o4-mini-deep-research: Text, Images
- Self-hostingSarvam-MPublishes downloadable weights (Apache 2.0)
| Measure | Weight | Gemini 2.5 Pro Preview 06-05 | Sarvam-M | o4-mini-deep-research |
|---|---|---|---|---|
| Inputs & features | 60% | 80 | 10 | 60 |
| Context window | 40% | 61 | 0 | 32 |
| Overall | 100% | 72/100 | 6/100 | 49/100 |
Left out because at least one model lacks the data: capability and price. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Gemini 2.5 Pro Preview 06-05Google | Sarvam-MSarvam AI | o4-mini-deep-researchOpenAI |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $1.22 (best) | — | $1.80 |
| Output | $9.75 | — | $7.20 (best) |
| Cached input | — | — | — |
| Blended (3:1) | $3.35 | — | $3.15 (best) |
| Long-context rate | Same rate | — | Same rate |
| Price source | Median of 4 providers | — | Median of 1 providers |
| Limits | |||
| Context window | 1,048,576 tokens (best) | 32,768 tokens | 200,000 tokens |
| Max output | 65,536 tokens | 32,768 tokens | 100,000 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | Yes |
| PDFs | Yes | No | No |
| Audio | Yes | No | No |
| Video | No | No | No |
| Reasoning | Yes | Yes | Yes |
| Tool calling | Yes | No | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Proprietary | OpenApache 2.0 | Proprietary |
| API model ID | — | — | — |
| API providers | 4 (best) | — | 1 |
| Released | Jun 5, 2025 | May 20, 2025 | Jun 26, 2025 |
| Knowledge cutoff | Jan 31, 2025 | — | May 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Gemini 2.5 Pro Preview 06-05$31.69
- Sarvam-M—
- o4-mini-deep-research$32.40
Which should you choose?
Which is better: Gemini 2.5 Pro Preview 06-05, Sarvam-M or o4-mini-deep-research?
Gemini 2.5 Pro Preview 06-05 is the better all-round choice, scoring 72/100 against o4-mini-deep-research (49) and Sarvam-M (6). It leads on inputs & features and context window. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Gemini 2.5 Pro Preview 06-05, Sarvam-M or o4-mini-deep-research?
o4-mini-deep-research is cheaper at $1.80 input / $7.20 output per million tokens (median across 1 API provider). Gemini 2.5 Pro Preview 06-05 costs $1.22 input / $9.75 output per million tokens (median across 4 API providers). At a typical mix of three input tokens to one output token, that is $3.15 per million tokens for o4-mini-deep-research versus $3.35 for Gemini 2.5 Pro Preview 06-05 (1.1× as much). Sarvam-M has no published per-token price.
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Gemini 2.5 Pro Preview 06-05 has not been scored yet, Sarvam-M has not been scored yet and o4-mini-deep-research has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Gemini 2.5 Pro Preview 06-05, Sarvam-M and o4-mini-deep-research yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Sarvam-M does not support tool calling, which most coding agents need.
Which has the bigger context window?
Gemini 2.5 Pro Preview 06-05 has the largest context window at 1,048,576 tokens, against 200,000 for o4-mini-deep-research and 32,768 for Sarvam-M. Maximum output per response: Gemini 2.5 Pro Preview 06-05 up to 65,536, Sarvam-M up to 32,768, o4-mini-deep-research up to 100,000 tokens.
Which can read images, PDFs, audio or video?
Gemini 2.5 Pro Preview 06-05 accepts text, images, PDFs and audio; Sarvam-M accepts text; o4-mini-deep-research accepts text and images. Gemini 2.5 Pro Preview 06-05 handles the widest range of inputs.
Are any of these open source?
Sarvam-M publishes its weights (Apache 2.0) and can be self-hosted; Gemini 2.5 Pro Preview 06-05 and o4-mini-deep-research is proprietary.
Which is newer?
o4-mini-deep-research is the newest, released Jun 26, 2025. Gemini 2.5 Pro Preview 06-05 came out Jun 5, 2025; Sarvam-M came out May 20, 2025. Knowledge cutoff: Gemini 2.5 Pro Preview 06-05 Jan 31, 2025, o4-mini-deep-research May 2024.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 2, 2026.
