Data as of November 4, 2026 · How the score is built
Released Nov 4, 20261M contextRicursive Ripple-65B Preview model card
Ripple-65B Preview
Decision readingRipple-65B Preview is tracked, but not publicly ranked yet. The profile exposes 17 sourced benchmark rows and leaves unsupported fields blank until a published record exists.
Released Nov 4, 2026 — see all recent releases
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Capability
Unranked
field median 50.5Not eligible for a public rank
Price
API rate not published
input median $0.75No comparable first-party hosted token rate
Speed
Not measured
field median 92 tok/sTime to first token not measured
Context
1Mtokens
field median 256,000Maximum output length is tracked separately
Strongest published evidence
Published rows are visible, but no category has enough eligible evidence for a comparative rank.
Validate before choosing
17 published rows leave some tracked benchmark slots empty. No comparable first-party API token rate is published.
Source-linked · 17 displayable benchmark rows
Category score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank Not rankedWeight 22%4 benchmarksVerified | 62.5 | 4 benchmarks | Verified | |||
| CodingRank Not rankedWeight 20%5 benchmarksVerified | 51.6 | 5 benchmarks | Verified | |||
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MultimodalWeight 12%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| KnowledgeRank Not rankedWeight 12%3 benchmarksVerified | 78.4 | 3 benchmarks | Verified | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingRank Not rankedWeight 5%1 benchmarkVerified | 86.2 | 1 benchmark | Verified | |||
| MathRank Not rankedWeight 5%4 benchmarksVerified | 84.5 | 4 benchmarks | Verified |
17 of 797 tracked benchmark slots have displayable evidence · bars run 0–100
Coverage detailsHow much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic4/4 verified
- Coding5/5 verified
- ReasoningNot measured
- MultimodalNot measured
- Knowledge3/3 verified
- MultilingualNot measured
- Inst. Following1/1 verified
- Math4/4 verified
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding5 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench Pro | Score61.4% | Versus best verified row Best verified: Claude Opus 5.5 · 89.9% | Gap28.5 behind | WeightDisplay only | Provider exact |
| DeepSWE | Score44.3% | Versus best verified row Best verified: Muse Wonder Full · 79.0% | Gap34.7 behind | WeightDisplay only | Provider exact |
| SciCodeScientific Code Benchmark | Score39.4% | Versus best verified row Best verified: Sakana Fugu · 60.1% | Gap20.7 behind | WeightDisplay only | Provider exact |
| Terminal-Bench 2.1 | Score80.5% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap12.3 behind | WeightDisplay only | Provider exact |
| SWE-bench VerifiedSoftware Engineering Benchmark Verified | Score76.5% | Versus best verified row Best verified: Claude Opus 5 · 96% | Gap19.5 behind | WeightDisplay only | Provider exact |
Agentic4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 | Score80.5% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap12.3 behind | WeightDisplay only | Provider exact |
| AutomationBench | Score36.4% | Versus best verified row Best verified: MiniMax M3.1 Flash · 62.2% | Gap25.8 behind | WeightDisplay only | Provider exact |
| MCP Atlas | Score71.7% | Versus best verified row Best verified: Muse Spark Open · 89.2% | Gap17.5 behind | WeightDisplay only | Provider exact |
| τ³-bench resultsτ³-Bench Tool-Agent-User Evaluation | Score31.3% | Versus best verified row Best verified: Mercury 2.5 · 96.0% | Gap64.7 behind | WeightDisplay only | Provider exact |
Knowledge3 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMLU-ProMassive Multitask Language Understanding Professional | Score84.2% | Versus best verified row Best verified: Qwen3.7 Max · 89.6% | Gap5.4 behind | WeightDisplay only | Provider exact |
| MMLUMassive Multitask Language Understanding | Score89.6% | Versus best verified row Best verified: o1 · 91.8% | Gap2.2 behind | WeightDisplay only | Provider exact |
| GPQA-DGPQA Diamond | Score85.2% | Versus best verified row Best verified: GPT-6 Astra · 96.0% | Gap10.8 behind | WeightDisplay only | Provider exact |
Inst. Following1 row
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| IFBenchInstruction Following Benchmark | Score79.2% | Versus best verified row Best verified: Flint · 89.2% | Gap10 behind | WeightDisplay only | Provider exact |
Math4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| AIME26AIME 2026 | Score95.0% | Versus best verified row Best verified: GLM-5.2 · 99.2% | Gap4.2 behind | WeightDisplay only | Provider exact |
| HMMT Feb 2026Harvard-MIT Mathematics Tournament February 2026 | Score90.5% | Versus best verified row Best verified: Qwen3.7 Max · 97.1% | Gap6.6 behind | WeightDisplay only | Provider exact |
| AIME 2024American Invitational Mathematics Examination 2024 | Score97.1% | Versus best verified row Best verified: Ripple-65B Preview · 97.1% | GapBest verified | WeightDisplay only | Provider exact |
| AIME 2025American Invitational Mathematics Examination 2025 | Score98.3% | Versus best verified row Best verified: GPT-5.2 · 100% | Gap1.7 behind | WeightDisplay only | Provider exact |
Bars run 0–100; the dark tick marks the best source-verified value
All 17 rowsLineage
The sequence follows explicit supersedes links. Each score is estimated for that model; a relative can inform a sparse estimate but never sets a floor, so a newer release can score below an earlier one. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
Base entry
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- ricursiveai/ripple-65b-previewOpenRouter Ripple-65B Preview route
- Context window
- 1MRicursive Ripple-65B Preview model card
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- textRicursive Ripple-65B Preview model card
- Output modalities
- textRicursive Ripple-65B Preview model card
- Parameters
- Not sourced yet
- Availability
- Hugging Face · self-hosted inference · OpenRouter (GMI Cloud)Ricursive: Introducing Ripple-65B Preview
- Cloud regions
- Not tracked yet
- Lifecycle
- activeRicursive Ripple-65B Preview model card
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing recordRicursive launch post
- Self-host
- Ripple-65B Preview needs ~80GB VRAM (1× NVIDIA H100 (80GB)).
- Rate limits
- Not tracked yet
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
We track Ripple-65B Preview, but the public leaderboard excludes this profile until enough non-generated benchmark coverage is available. Only published rows appear above.
Ripple-65B Preview is a open weight model with a 1M context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Open Apache 2.0 BF16 weights on Hugging Face (ricursiveAI/Ripple-65B-Preview), with an NVFP4 checkpoint, a DFlash2 drafter and a BF16 base model alongside. Ricursive has no first-party API. At launch OpenRouter routes ricursiveai/ripple-65b-preview to GMI Cloud in BF16 at $0.20 input, $0.05 cached input and $0.60 output per million tokens with a 16,384-token context; Together AI lists it as coming soon and Ricursive says Fireworks follows. Ricursive reports 250 tokens per second on two B200 GPUs at batch 8 before quantization.
Ricursive Intelligence, the chip-design AI lab behind AlphaChip (ricursive.com, not the Recursive lab that released Flint), released Ripple-65B Preview as open weights on November 4, 2026 under Apache 2.0. It is a 65B-parameter dense, text-only Transformer pretrained from scratch on 12T predominantly English and code tokens, with an 80-layer, 5,120-wide thin-and-deep design, gated attention, global NoPE, value residuals and a 1M-token context window. Ricursive calls it a general-purpose variant of the proprietary model behind its chip design platform, with light post-training focused on agentic coding, and says the full program cost $32M. Benchmark values are the instruct model card's Ripple column at full BF16 precision and maximum effort: OpenCode where no harness is named and Terminus 2 for DeepSWE 1.1. Terminal-Bench 2.1 is 80.5 both self-reported and with Terminus 2, τ³-bench is the banking domain, SWE-bench Pro is v1.0 and AutomationBench is the public set. SWE-bench Pro v2.0 Hard (79.1), FORTRESS Adversarial and Benign (83.8, 95.5) and StrongREJECT (97.5) have no compatible lane and stay unmapped, as does the pretrain-only base model table. All rows are provider-run, so the model stays display-only and outside rankings.
17 of 797 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Last updated November 4, 2026. Runtime fields remain blank until a sourced snapshot exists.
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Questions
How does Ripple-65B Preview perform overall in AI benchmarks?
Ripple-65B Preview has 17 source-displayable benchmark rows, but it does not qualify for a public overall rank. The available rows remain visible by category without being converted into a site-wide score. Missing evidence stays blank instead of being estimated from an earlier model.
Is Ripple-65B Preview good for knowledge and understanding?
Ripple-65B Preview has source-displayable benchmark coverage for knowledge and understanding, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Ripple-65B Preview good for coding and programming?
Ripple-65B Preview has source-displayable benchmark coverage for coding and programming, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Ripple-65B Preview good for mathematics?
Ripple-65B Preview has source-displayable benchmark coverage for mathematics, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Ripple-65B Preview good for agentic tool use and computer tasks?
Ripple-65B Preview has source-displayable benchmark coverage for agentic tool use and computer tasks, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Ripple-65B Preview good for instruction following?
Ripple-65B Preview has source-displayable benchmark coverage for instruction following, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Ripple-65B Preview open source?
Ripple-65B Preview is an open-weight model from Ricursive Intelligence. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Does Ripple-65B Preview have full benchmark coverage on BenchLM?
No. Ripple-65B Preview currently has 18 source-displayable rows. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of Ripple-65B Preview?
Ripple-65B Preview has a documented context window of 1M. That figure is the maximum combined prompt and retained-conversation space reported for this exact model; it is not the maximum output length. The profile keeps output limits separate because providers often publish those limits independently.