Command A vs Sonar Deep Research vs Step 1 (32K)
Command A comes out ahead, 24 to 20 and 18 on our weighted score, though Sonar Deep Research is 20% cheaper per token.
- Our pick
CohereCommand A
24/100- ECI—
- Price$2.50 / $10.00
- Context256K
PerplexitySonar Deep Research
20/100- ECI—
- Price$2.00 / $8.00
- Context128K
StepFunStep 1 (32K)
18/100- ECI—
- Price$2.05 / $9.59
- Context33K
Command A is our pick
Command A is the better all-round choice, scoring 24/100 against Sonar Deep Research (20) and Step 1 (32K) (18). It leads on context window. Sonar Deep Research wins on price. 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 priceSonar Deep ResearchSonar Deep Research $3.50 · Step 1 (32K) $3.94 · Command A $4.38 per 1M tokens (3:1 blend)
- Longest contextCommand ACommand A 256,000 · Sonar Deep Research 128,000 · Step 1 (32K) 32,768 tokens
- Widest inputsSame inputsCommand A: Text · Sonar Deep Research: Text · Step 1 (32K): Text
- Self-hostingCommand APublishes downloadable weights
| Measure | Weight | Command A | Sonar Deep Research | Step 1 (32K) |
|---|---|---|---|---|
| Price | 50% | 19 | 24 | 22 |
| Inputs & features | 30% | 25 | 10 | 25 |
| Context window | 20% | 36 | 24 | 0 |
| Overall | 100% | 24/100 | 20/100 | 18/100 |
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 | Command ACohere | Sonar Deep ResearchPerplexity | Step 1 (32K)StepFun |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $2.50 | $2.00 (best) | $2.05 |
| Output | $10.00 | $8.00 (best) | $9.59 |
| Cached input | — | — | $0.41 |
| Blended (3:1) | $4.38 | $3.50 (best) | $3.94 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Cohere API | Official Perplexity API | Official StepFun (Global) API |
| Limits | |||
| Context window | 256,000 tokens (best) | 128,000 tokens | 32,768 tokens |
| Max output | 8,000 tokens | 32,768 tokens (best) | 32,768 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | Yesminimal · low · medium · high | No |
| Tool calling | Yes | No | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | Proprietary | Proprietary |
| API model ID | command-a-03-2025 | sonar-deep-research | step-1-32k |
| API providers | 3 (best) | 3 (best) | 1 |
| Released | Mar 13, 2025 | Feb 1, 2025 | Jan 1, 2025 |
| Knowledge cutoff | Jun 1, 2024 | Jan 2025 | Jun 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Command A$45.00
- Sonar Deep Research$36.00
- Step 1 (32K)$39.68
Which should you choose?
Which is better: Command A, Sonar Deep Research or Step 1 (32K)?
Command A is the better all-round choice, scoring 24/100 against Sonar Deep Research (20) and Step 1 (32K) (18). It leads on context window. Sonar Deep Research wins on price. 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, Command A, Sonar Deep Research or Step 1 (32K)?
Sonar Deep Research is cheaper at $2.00 input / $8.00 output per million tokens (official Perplexity API price). Step 1 (32K) costs $2.05 input / $9.59 output per million tokens (official StepFun (Global) API price); Command A costs $2.50 input / $10.00 output per million tokens (official Cohere API price). At a typical mix of three input tokens to one output token, that is $3.50 per million tokens for Sonar Deep Research versus $3.94 for Step 1 (32K) (1.1× as much) and $4.38 for Command A (1.3× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Command A has not been scored yet, Sonar Deep Research has not been scored yet and Step 1 (32K) has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Command A, Sonar Deep Research and Step 1 (32K) yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Sonar Deep Research does not support tool calling, which most coding agents need.
Which has the bigger context window?
Command A has the largest context window at 256,000 tokens, against 128,000 for Sonar Deep Research and 32,768 for Step 1 (32K). Maximum output per response: Command A up to 8,000, Sonar Deep Research up to 32,768, Step 1 (32K) up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Command A accepts text; Sonar Deep Research accepts text; Step 1 (32K) accepts text. They handle the same number of input types.
Are any of these open source?
Command A publishes its weights and can be self-hosted; Sonar Deep Research and Step 1 (32K) is proprietary.
Which is newer?
Command A is the newest, released Mar 13, 2025. Sonar Deep Research came out Feb 1, 2025; Step 1 (32K) came out Jan 1, 2025. Knowledge cutoff: Command A Jun 1, 2024, Sonar Deep Research Jan 2025, Step 1 (32K) Jun 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 Oct 4, 2026.
