Model
Model explorer

ERNIE 5.0 (Preview)

CLOSED
Baidu · ERNIE family · released Nov 13, 2025

2.4T-parameter natively omni-modal flagship unveiled at Baidu World 2025; scored 22 on Artificial Analysis's Intelligence Index as a free preview.

ReasoningCodingVisionFunction callingTool useAgentic
1681.9
Elo · rank #126
Parameters
2400B
Active params
Undisclosed
Context
128K tokens
Architecture
Natively omni-modal Mixture-of-Experts, unified autoregressive next-token objective, <3% of experts active per token
License
Proprietary
Languages
—
API price (in/out)
No hosted API
Modalities
text · vision · audio · video
Benchmark results
Bar shows position within the tracked field; marker = field best
Arena EloHuman preference1451#47
best: Muse Wonder · 1513
GPQA DiamondReasoning78.0%#103
best: GPT-6 Astra · 96.0%
Humanity's Last ExamReasoning13.0%#97
best: Kimi K3.1 Pro · 70.0%
IFBenchReasoning41.0%#53
best: MiniMax M3 · 83.0%
MMMU-ProVision65.0%#43
best: Claude Opus 4.7 · 85.5%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
—
Output / M tok
—
API price — · each benchmark row carries its own source badge (see methodology)