Mercury 2 vs Sarvam 105B vs Qwen3.5 9B Base
Too close to call on our weighted score (Sarvam 105B 52, Mercury 2 52, Qwen3.5 9B Base 50). The right pick depends on what you value most.
InceptionMercury 2
52/100- ECI—
- Price$0.25 / $0.75
- Context128K
Sarvam AI
Sarvam 105B
52/100- ECI—
- Price$0.158 / $0.632
- Context131K
Alibaba (Qwen)Qwen3.5 9B Base
50/100- ECI—
- Price$0.30 / $0.807
- Context262K
Too close to call
It is close. Our weighted score puts them within a point (Sarvam 105B 52/100, Mercury 2 52/100, Qwen3.5 9B Base 50/100), so choose by what matters most for your work: Sarvam 105B on price and Qwen3.5 9B Base for long inputs. The score weighs price 50%, inputs & features 30%, context window 20%. 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 priceSarvam 105BSarvam 105B $0.276 · Mercury 2 $0.375 · Qwen3.5 9B Base $0.427 per 1M tokens (3:1 blend)
- Longest contextQwen3.5 9B BaseQwen3.5 9B Base 262,144 · Sarvam 105B 131,072 · Mercury 2 128,000 tokens
- Widest inputsQwen3.5 9B BaseMercury 2: Text · Sarvam 105B: Text · Qwen3.5 9B Base: Text, Images, Video
- Self-hostingSarvam 105B and Qwen3.5 9B BasePublishes downloadable weights (Apache-2.0)
| Measure | Weight | Mercury 2 | Sarvam 105B | Qwen3.5 9B Base |
|---|---|---|---|---|
| Price | 50% | 67 | 73 | 65 |
| Inputs & features | 30% | 45 | 35 | 35 |
| Context window | 20% | 24 | 25 | 37 |
| Overall | 100% | 52/100 | 52/100 | 50/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Mercury 2Inception | Sarvam 105BSarvam AI | Qwen3.5 9B BaseAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.25 | $0.158 (best) | $0.30 |
| Output | $0.75 | $0.632 (best) | $0.807 |
| Cached input | $0.025 (best) | $0.053 | — |
| Blended (3:1) | $0.375 | $0.276 (best) | $0.427 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Inception API | Official Sarvam AI API | Median of 1 providers |
| Limits | |||
| Context window | 128,000 tokens | 131,072 tokens | 262,144 tokens (best) |
| Max output | 50,000 tokens | 131,072 tokens (best) | 65,536 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | Yes |
| Reasoning | Yeslow · medium · high | Yeslow · medium · high | No |
| Tool calling | Yes | Yes | No |
| Structured output | Yes | No | No |
| Availability | |||
| Weights | Proprietary | Open | OpenApache-2.0 |
| API model ID | mercury-2 | sarvam-105b | — |
| API providers | 7 (best) | 3 | 1 |
| Released | Feb 24, 2026 | Feb 18, 2026 | Feb 26, 2026 |
| Knowledge cutoff | — | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Mercury 2$4.00
- Sarvam 105B$2.84
- Qwen3.5 9B Base$4.61
Which should you choose?
Which is better: Mercury 2, Sarvam 105B or Qwen3.5 9B Base?
It is close. Our weighted score puts them within a point (Sarvam 105B 52/100, Mercury 2 52/100, Qwen3.5 9B Base 50/100), so choose by what matters most for your work: Sarvam 105B on price and Qwen3.5 9B Base for long inputs. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Mercury 2, Sarvam 105B or Qwen3.5 9B Base?
Sarvam 105B is cheaper at $0.158 input / $0.632 output per million tokens (official Sarvam AI API price). Mercury 2 costs $0.25 input / $0.75 output per million tokens (official Inception API price); Qwen3.5 9B Base costs $0.30 input / $0.807 output per million tokens (median across 1 API provider). At a typical mix of three input tokens to one output token, that is $0.276 per million tokens for Sarvam 105B versus $0.375 for Mercury 2 (1.4× as much) and $0.427 for Qwen3.5 9B Base (1.5× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Mercury 2 has not been scored yet, Sarvam 105B has not been scored yet and Qwen3.5 9B Base has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Mercury 2, Sarvam 105B and Qwen3.5 9B Base yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Qwen3.5 9B Base does not support tool calling, which most coding agents need.
Which has the bigger context window?
Qwen3.5 9B Base has the largest context window at 262,144 tokens, against 131,072 for Sarvam 105B and 128,000 for Mercury 2. Maximum output per response: Mercury 2 up to 50,000, Sarvam 105B up to 131,072, Qwen3.5 9B Base up to 65,536 tokens.
Which can read images, PDFs, audio or video?
Mercury 2 accepts text; Sarvam 105B accepts text; Qwen3.5 9B Base accepts text, images and video. Qwen3.5 9B Base handles the widest range of inputs.
Are any of these open source?
Sarvam 105B and Qwen3.5 9B Base publishes its weights (Apache-2.0) and can be self-hosted; Mercury 2 is proprietary.
Which is newer?
Qwen3.5 9B Base is the newest, released Feb 26, 2026. Mercury 2 came out Feb 24, 2026; Sarvam 105B came out Feb 18, 2026.
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.
