Language modelOpen weights

Naive-N0.5-Flash

NaiveAI

Released: -
Data date: October 3, 2026

Naive-N0.5-Flash combines sliding-window attention with DeepSeek sparse attention for a 1M-token context window. The open MoE checkpoint targets coding and AI research, activating 15.5 of its 309 billion parameters per token.

NaiveAI releases the weights under MIT. Its model card documents FP8 deployment on NVIDIA hardware.

Specifications and access

Specification	Value and source
Model ID	NaiveAI/Naive-N0.5-FlashSource
Model class	Mixture-of-experts model for coding and AI researchSource
Parameters	309 billion total, 15.5 billion activeSource
Context window	1M tokens according to the model cardSource
Architecture	Hybrid of sliding-window attention and DeepSeek sparse attentionSource
License	MITSource

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.

MLE-bench-30 (Show measurement, test conditions, and source)

Source value: 73.7
Score: 73.7
Metric: MLE-bench-30
Unit: %
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Provider model card

DeepSWE v1.1 (Show measurement, test conditions, and source)

Source value: 67.8
Score: 67.8
Metric: DeepSWE v1.1 reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

Agents Last Exam (Show measurement, test conditions, and source)

Source value: 32.4
Score: 32.4
Metric: Agents Last Exam reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

TerminalBench 2.1 (Show measurement, test conditions, and source)

Source value: 86.7
Score: 86.7
Metric: TerminalBench 2.1 reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

SWE-Bench Pro (Show measurement, test conditions, and source)

Source value: 73.6
Score: 73.6
Metric: SWE-Bench Pro reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

ProgramBench (Show measurement, test conditions, and source)

Source value: 17.5
Score: 17.5
Metric: ProgramBench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

NL2Repo (Show measurement, test conditions, and source)

Source value: 71.9
Score: 71.9
Metric: NL2Repo reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

FrontierSWE v1 (Show measurement, test conditions, and source)

Source value: 78.2
Score: 78.2
Metric: FrontierSWE v1 reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

PostTrainBench (Show measurement, test conditions, and source)

Source value: 37.5
Score: 37.5
Metric: PostTrainBench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

PaperBench (Show measurement, test conditions, and source)

Source value: 63.2
Score: 63.2
Metric: PaperBench reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

SOL-ExecBench meanSOL (Show measurement, test conditions, and source)

Source value: 72.81
Score: 72.81
Metric: SOL-ExecBench meanSOL reported score
Unit: No unit provided
Category: source-specific
Direction: Higher is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

NanoChat AutoResearch validation BPB (Show measurement, test conditions, and source)

Source value: 0.9051
Score: 0.9051
Metric: NanoChat AutoResearch validation BPB reported score
Unit: No unit provided
Category: source-specific
Direction: Lower is better
Source type: vendor-reported
Evaluator: Provider model card
Status: active
Retrieved at: 2026-10-04
Methodology: Vendor-reported result. Official image-only card, visually inspected. Default Claude Code 2.1.207, 1M context, tool environment; proprietary agent/R&D harnesses.
Context: Source-specific observation; it is not a shared comparison cohort. The source table does not state a unit or scale for this numeric score.

Source Provider model card

Sources and data date

Every statement links to its underlying documentation or leaderboard.

Type	Evidence and data date
Research status	Research date October 4, 2026. 1 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Additional source	Naive-N0.5-Flash model card (retrieved October 3, 2026; October 4, 2026) · Naive-N0.5-Flash model card · Editorial description reviewed October 4, 2026