# Nemotron 3.5 Bolt E4B

NVIDIA · Open Weight · bench-align-v5

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/models/nemotron-3-5-bolt-e4b  
JSON: https://modelchoice.dev/api/model/nemotron-3-5-bolt-e4b

## Facts

| Field | Value | Evidence |
| --- | --- | --- |
| Canonical id | `nemotron-3-5-bolt-e4b` | — |
| Overall score | Unavailable | Unavailable · source benchlm:models |
| Verified score (BenchLM): a separate BenchLM figure, not the overall score | 40 | Observed 2026-11-21 · source benchlm:models |
| Context window | 64K tokens | Observed 2026-11-21 · source benchlm:models |
| Release date | 2026-10-23 | Observed 2026-11-21 · source benchlm:models |
| Access type | Open Weight | — |
| Blended $/1M (75% input / 25% output) | Unavailable | Derived from the input and output rates below |

**Only a verified score is published.** BenchLM publishes no overall score for this model, so the overall score stays unavailable. The verified score beside it is a separate figure in BenchLM's data. BenchLM does not document how the two are calculated differently, and they can differ widely, so one is never used in place of the other.

## Capability evidence

Seven axes from the ranking source. An axis the source did not score is unavailable, not zero. Beside each score is BenchLM's separate verified score for that axis; the two are published side by side and are not interchangeable.

| Axis | Score | Evidence | Verified (BenchLM) | Verified evidence |
| --- | --- | --- | --- | --- |
| Agentic | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Coding | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Knowledge | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Reasoning | 23.4 | Observed 2026-11-21 · source benchlm:models | 23.4 | Observed 2026-11-21 · source benchlm:models |
| Multimodal & Grounded | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Instruction Following | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Math | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |

## Runtime service evidence

Measured values with the date they were observed. Nothing is inferred from a sibling model or a provider claim. Regional or per-endpoint measurements appear only when the API supplies them; none are modelled.

| Measurement | Value | Observed | Last good | Evidence |
| --- | --- | --- | --- | --- |
| Time to first token | Unavailable | Unavailable | Unavailable | Unavailable · source benchlm:speed |
| Throughput | Unavailable | Unavailable | Unavailable | Unavailable · source benchlm:speed |

## Endpoint and price matrix

Every published price component, including cache reads and writes.

| Component | USD | Evidence |
| --- | --- | --- |
| Input / 1M tokens | Unavailable | Unavailable · source benchlm:pricing |
| Output / 1M tokens | Unavailable | Unavailable · source benchlm:pricing |
| Cache read / 1M tokens | Unavailable | Unavailable · source benchlm:pricing |
| Cache write / 1M tokens | Unavailable | Unavailable · source openrouter:pricing |
| Blended / 1M (75% input / 25% output) | Unavailable | Derived — from the input and output rates above; it has no source record of its own |
| Cost per successful task (LiveBench) | Unavailable | Unavailable · source livebench:table |

**Pricing note.** NVIDIA publishes downloadable OpenMDW 1.1 Q4_K_M weights for local inference and lists the model on build.nvidia.com without a per-token tariff. No first-party hosted API price is established for this checkpoint; hardware costs are separate.

## Workload-aware monthly cost example

10 conversations per day × 8 messages × 22 active days, 1200 input and 400 output tokens per message, no cache. Derived here from the published rates above by this site's own calculator — not a figure any source published.

| Field | Value |
| --- | --- |
| Modelled monthly cost | Unavailable |
| Modelled tokens | Unavailable |
| Reason | The applicable input rate is unavailable. |

## Benchmark record

21 matched benchmark rows with their published value, unit, and provenance.

### Reasoning

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| LongBench v2 | 48.67 | Long-context tasks | Hard long-context | [LongBench v2](https://arxiv.org/abs/2412.15204) |
| RULER (64K) (RULER at a 64K-token context) | 85.76 | Synthetic long-context tasks | Long-context retrieval and reasoning | [RULER: What's the Real Context Size of Your Long-Context Language Models?](https://github.com/NVIDIA/RULER) |
| LongMemEval | 69.92 | Questions over multi-session chat histories | Recall across conversations | [LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory](https://github.com/xiaowu0162/LongMemEval) |

### Instruction Following

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| AA-IFBench (Artificial Analysis IFBench) | 61.1 | Verifiable instruction constraints | Instruction precision | [Artificial Analysis IFBench Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/ifbench) |
| MultiChallenge (provider reports) (MultiChallenge: provider-reported runs) | 55.60 | — | — | [MultiChallenge](https://labs.scale.com/leaderboard/multichallenge) |
| Brevity Benchmark (Brevity Benchmark (NVIDIA)) | 82.84 | Conversational prompts (not yet published) | Concise, speech-ready answers | [NVIDIA Nemotron 3.5 Bolt E4B 4GB](https://research.nvidia.com/labs/nemotron/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB/) |
| EQ-Bench 3 | 38.15 | Multi-turn roleplay scenarios | Emotional and social intelligence | [EQ-Bench 3](https://eqbench.com/eqbench3.html) |
| RoleBench (RoleBench (RoleLLM)) | 71.21 | Character role-play prompts | Character speaking style | [RoleLLM: Benchmarking, Eliciting, and Enhancing Role-Playing Abilities of Large Language Models](https://github.com/InteractiveNLP-Team/RoleLLM-public) |
| RoleMRC | 79.12 | Role-play and in-role instruction prompts | Role identity adherence | [RoleMRC](https://github.com/LuJunru/RoleMRC) |
| IHEval (IHEval: instruction hierarchy evaluation) | 82.40 | Aligned and conflicting instruction sets | Instruction priority | [IHEval: Evaluating Language Models on Following the Instruction Hierarchy](https://github.com/ytyz1307zzh/IHEval) |
| AgentIF (AgentIF: instruction following in agentic scenarios) | 22.95 | Agentic system prompts and instructions | Complex agentic instructions | [AgentIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios](https://agentif.github.io/) |
| InverseIFEval | 45.26 | Counterintuitive instructions | Overriding trained conventions | [InverseIFEval](https://arxiv.org/abs/2509.04292) |
| StructEval-T (StructEval text-format subset) | 79.99 | Structured-output generation and conversion | Valid structured output | [StructEval](https://structeval.github.io/) |
| SysBench (SysBench: system message following) | 67.22 | System-message constraints over conversations | Keeping long prompt rules | [SysBench: Can Large Language Models Follow System Messages?](https://github.com/PKU-Baichuan-MLSystemLab/SysBench) |

### Agentic

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| τ²-bench results (τ²-Bench Tool-Agent-User Evaluation) | 52.32 | Airline, retail, and telecom customer-service task sets | Dual-control customer-service workflows | [τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment](https://arxiv.org/abs/2506.07982) |
| BFCL v3 (Berkeley Function Calling Leaderboard v3) | 67.3 | Function-calling tasks | Advanced tool use | [The Berkeley Function Calling Leaderboard (BFCL): From Tool Use to Agentic Evaluation of Large Language Models](https://gorilla.cs.berkeley.edu/leaderboard.html) |
| When2Call Multi-Turn (When2Call Multi-Turn (NVIDIA)) | 60.50 | Multi-turn tool-use decisions (not yet published) | When to call, clarify or abstain | [NVIDIA Nemotron 3.5 Bolt E4B 4GB](https://research.nvidia.com/labs/nemotron/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB/) |
| ToolSandbox | 74.13 | Stateful tool-use conversations | Tool use with changing state | [ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities](https://github.com/apple-aiml-research/ToolSandbox) |
| RelyToolBench | 69.29 | Tool-use requests | Tool grounding and hallucination | [RelyToolBench](https://github.com/X-LANCE/ToolHallucination) |

### Knowledge

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| GaRAGe | 40.73 | Questions with retrieved passages | Facts from supplied sources | [GaRAGe](https://github.com/amazon-science/GaRAGe) |
| RAGTruth (provider reports) (RAGTruth: provider-reported F1) | 77.88 | Retrieval-augmented question answering, summarization and data-to-text | Avoiding unsupported claims | [RAGTruth](https://github.com/ParticleMedia/RAGTruth) |

## Lifecycle and limitations log

Lifecycle events the source associates with this model.

No lifecycle event references this model. That is not evidence the model has no lifecycle plan — only that this source published none.

## What this profile does not claim

Values are reproduced exactly as their sources published them, in the units those sources declared; none are converted, interpolated, or averaged across providers. Any field marked unavailable was attempted and not returned. Last attempted fetch for this model's score: 2026-11-21 06:17 UTC.
