Cohere Rerank 3.5
Cohere Rerank 3.5 is a previous Cohere rerank snapshot, not the live docs lead. Rechecked 2026-08-17: docs.cohere.com/docs/rerank now leads with rerank-v4.0 and rerank-v4.0-pro. The same family still lists leftover English / multilingual v3.0 rows. Drop leftover industry-leading / 100+ languages as 2026 default / millisecond inference / 4096 as a crown / BEIR +30% nDCG / $2 per 1,000 calls.
Compare Voyage Rerank 2 if you wanted another leftover rerank SKU (live Voyage docs lead with rerank-2.5 at $0.05 / $0.02), and Jina AI Reranker v3 if you wanted a leftover Jina card (live page leads with reranker-v3.5).
Key Features
- Previous 3.5 snapshot: Leftover copy treated 3.5 as the current Cohere default. Recheck the rerank docs before you hard-code it.
- Not the 2026 rerank lead: Live docs lead with rerank-v4.0 / rerank-v4.0-pro.
- No invented quality/latency: Do not paste leftover BEIR +30% nDCG or millisecond inference unless you re-run a first-party card.
- No leftover $2 / 1k row: Recheck the live Cohere pricing or docs page before you invent a per-call dollar figure.
Use Cases
- People with a Rerank 3.5 bookmark who need the catalog corrected.
- People comparing leftover 3.5 vs live rerank-v4.0.
- People who were about to paste $2 / 1k and BEIR +30% into a 2026 deck.
Limitation: We did not call rerank-v4.0 or re-run BEIR.
Pricing
First-party pages on 2026-08-17.
| Piece | Price | Notes |
|---|---|---|
| rerank-3.5 | Previous snapshot | Recheck docs.cohere.com/docs/rerank. |
| Live docs lead | Recheck rerank-v4.0 | rerank-v4.0 / rerank-v4.0-pro lead the page. |
| Per-call dollar row | Recheck live pricing | Do not keep leftover $2 / 1,000 calls as a 2026 census. |
If a leftover blog still calls 3.5 the current Cohere default, treat the v4.0 docs as the correction.
Getting Started
- Open docs.cohere.com/docs/rerank before you hard-code
rerank-3.5. - Recheck
rerank-v4.0if you wanted the live Cohere path. - Recheck Voyage Rerank 2 if you wanted another leftover rerank SKU with a live 2.5 successor.
- Do not paste BEIR +30% or $2 / 1k into a deck.
Frequently Asked Questions
Is Rerank 3.5 still the Cohere default?
No. The docs we opened lead with rerank-v4.0.
Is it still $2 per 1,000 calls?
Do not keep that leftover row. Recheck the live Cohere pricing or docs page.
Same as Voyage Rerank 2?
No. Voyage Rerank 2 is a leftover Voyage SKU. Live Voyage docs lead with rerank-2.5.
Alternatives
- Voyage Rerank 2: Previous Voyage rerank snapshot; live docs lead with rerank-2.5.
- Jina AI Reranker v3: Previous Jina snapshot; live page leads with reranker-v3.5.
- BGE-M3: 2024 open multilingual embedding card.
Tips
- Call Rerank 3.5 a previous snapshot, not the current Cohere default.
- Point new Cohere rerank work at rerank-v4.0.
- Recheck live docs before you invent a per-call price.
Conclusion
Cohere Rerank 3.5 is a previous snapshot, not the 2026 Cohere rerank default. Start at docs.cohere.com/docs/rerank, then decide whether rerank-v4.0 already covers the path you need.
Comments
No comments yet. Be the first to comment!
Related Tools
rerank-multilingual-v3.0
docs.cohere.com/docs/rerank
Previous Cohere multilingual rerank snapshot. Rechecked: live docs lead with rerank-v4.0. Drop leftover 4096 / current-non-English crown.
Cohere: rerank-v3.5
docs.cohere.com/docs/rerank
Previous Cohere rerank-v3.5 snapshot. Rechecked: live docs lead with rerank-v4.0. Drop leftover SoTA / 4096 crown.
rerank-english-v3.0
docs.cohere.com/docs/rerank
Previous Cohere English rerank snapshot. Rechecked: live docs lead with rerank-v4.0. Drop leftover 4096 / current-English crown.
Related Insights
Hook OpenCode Go into Codex on Windows. Do Not Open a Second Toolkit.
A ChatGPT-signed Codex desktop app still shows mostly GPT in the picker. On Windows, enable only OpenCode Go and the Grok, GLM, Kimi, DeepSeek, and MiniMax models you already pay for appear in the same selector. Keys stay local. Native GPT stays put.
What Locks Codex Is the Picker, Not the Models
You already pay for OpenCode Go, Grok, and Z.ai, but the Codex picker still shows mostly GPT. The community project codex-router does not teach another install ritual. It puts subscriptions you already bought back into the selector, keeps keys on the machine, and leaves native GPT alone.
Seven AI Coding CLIs, Six Months: No Matter How Strong the Model, Work Needs Supervision
Claude Code, Codex, opencode, pi, omp and DeepSeek Harness all have personalities. After six months of deep use I run a division of labor: pi for the fastest cheapest reviews, omp for complex PRs, DeepSeek Harness on V4 Flash for high-frequency low-cost review, and Claude Code, Qoder and Cursor for writing. No matter how strong the model, work needs supervision — ideally from an independent third party.