Sonar Deep Research
Unavailable in this round
Language modelActive
Sonar Deep Research
Perplexity
- Released
- February 2025
- Data date
- September 8, 2026
- Price
- Input · Output / 1M tokens
Sonar Deep Research stands out through adjustable research effort. Its documented configuration offers low, medium, and high levels, with medium as the default. Perplexity aims it at web research, source synthesis, and cited reports.
Earlier Sonar documentation described synchronous and asynchronous research jobs. Perplexity's current migration page says asynchronous Sonar requests are no longer supported and points to background mode instead.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | ProprietarySource |
| Access | APISource |
| Context window | 128,000 tokensSource |
| Notes | Perplexity additionally charges $2 per 1M citation tokens, $3 per 1M reasoning tokens and $5 per 1,000 search queries; the Sonar Chat Completions API is replaced by the Agent API on September 27, 2026.Source |
| Capabilities | Researches the web and synthesizes sources into detailed reports with citations. (retrieved 2026-09-22)Source |
| Reasoning effort | low, medium, high (medium is the default) (retrieved 2026-09-22)Source |
| API interfaces | Synchronous and asynchronous requests. Long research jobs can be submitted and retrieved later. (retrieved 2026-09-22)Source |
| Release | 2025-02 (Only the month is documented, not an exact day)Source |
| API model ID | sonar-deep-researchSource |
| Reasoning mode | YesSource |
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
BrowseComp (Perplexity migration comparison) (Show measurement, test conditions, and source)
- Source value
- 29.1
- Score
- 29.1
- Metric
- accuracy
- Unit
- %
- Benchmark version
- Perplexity Sonar-to-Agent-API comparison, evaluation date not stated
- Category
- agentic
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Perplexity migration benchmarks
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. Publisher comparison of Sonar Chat Completions; api=sonar, label=Deep Research. BrowseComp accuracy.
- Context
- Historical Sonar baseline, not an Agent API preset or a fresh endpoint test. Evaluation date, sample count and inference settings are not disclosed.
DeepSearchQA (Perplexity migration comparison) (Show measurement, test conditions, and source)
- Source value
- 48.94
- Score
- 48.94
- Metric
- F1
- Unit
- %
- Benchmark version
- Perplexity Sonar-to-Agent-API comparison, evaluation date not stated
- Category
- agentic
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Perplexity migration benchmarks
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Vendor-reported result. Publisher comparison of Sonar Chat Completions; api=sonar, label=Deep Research. DeepSearchQA F1.
- Context
- Historical Sonar baseline, not an Agent API preset or a fresh endpoint test. Evaluation date, sample count and inference settings are not disclosed.
Our tests
SVG short test
One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.
Earlier tests
Pelican on a bicycle
A community classic for free-form SVG drawing.
Sonar Deep Research
API request failed
Reasoning (requested): high
Current attempt
- Run date
- Oct 2, 2026
- API provider
- openrouter
- Requested API model
- perplexity/sonar-deep-research
- Reasoning setting
- high
- Pinned endpoint
- Perplexity | perplexity/sonar-deep-research
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Failed
Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.
Earlier attempts (1)
- Oct 1, 2026API request failedReasoning setting: high
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)Popularity
27.8 / 100
Sources and data date
| Type | Evidence and data date |
|---|---|
| Price | Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. |
| Research status | Research date September 8, 2026. 8 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | parameters-billions Original wording No exact value located in the checked primary model card/API documentation. |
| Not found in the inspected sources | active-parameters-billions Original wording No exact value located in the checked primary model card/API documentation. |
| Not found in the inspected sources | knowledge-cutoff Original wording No exact value located in the checked primary model card/API documentation. |
| Unresolved | unattributed-product-mode-benchmarks Note The Deep Research product launch reports HLE accuracy 21.1% and SimpleQA accuracy 93.9%, but does not identify the evaluated API checkpoint. These product-mode values remain unattributed; the separate, explicitly named Sonar API migration results are recorded. |
| Unresolved | AA benchmark rows Original wording No exact Artificial Analysis model page located in the checked public sitemap. |
| Additional source | Perplexity migration benchmarks (retrieved October 15, 2026) |
| Additional source | Model metadata (retrieved September 8, 2026; September 22, 2026) · Perplexity Sonar Deep Research documentation · Editorial description reviewed October 2, 2026 |
| Additional source | Perplexity migration from Sonar to Agent API · Editorial description reviewed October 2, 2026 |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Creative Commons Attribution 4.0 |