ERNIE 5.0 Thinking Preview vs Gemini 2.5 Pro
Wins 1 of 7 areas
Agents
Wins 5 of 7 areas
Coding · Reasoning · Images and charts · Long documents · Following instructions
Gemini 2.5 Pro is the stronger all-rounder.ERNIE 5.0 Thinking Preview is cheaper and better at agents.
Scores updated · 19 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking03Gemini 2.5 Pro3 of 3 tests
- CodingWriting and fixing software02Gemini 2.5 Pro2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video01Gemini 2.5 Pro1 of 1 test
- Long documentsFinding answers in very long texts01Gemini 2.5 Pro1 of 1 test
- Following instructionsDoing exactly what it is asked01Gemini 2.5 Pro1 of 1 test
- AgentsCarrying out multi-step tasks on its own10ERNIE 5.0 Thinking Preview1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Agents, images and charts, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
ERNIE 5.0 Thinking Preview costs 60% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Gemini 2.5 Pro pulls ahead
- Reasons across sets of long documentsAA-LCR+62points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+17points ahead
- Harder college exam questions with imagesMMMU-Pro+10.3points ahead
Where ERNIE 5.0 Thinking Preview pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+3.9points ahead
Every test, side by side
All 19 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 2.5 Pro
- SciCodeGemini 2.5 Pro by 8.837.546.3+8.8
- Terminal-Bench HardGemini 2.5 Pro by 1.52526.5+1.5
AgentsERNIE 5.0 Thinking Preview
- GDPValERNIE 5.0 Thinking Preview by 14.414.40+14.4
ReasoningGemini 2.5 Pro
- GPQA DiamondGemini 2.5 Pro by 5.977.783.6+5.9
- Humanity's Last ExamGemini 2.5 Pro by 4.713.318+4.7
- CritPtGemini 2.5 Pro by 1.21.42.6+1.2
FactsEven
- AA-Omniscience · AccuracyGemini 2.5 Pro by 1722.139+17
- AA-Omniscience · Non-hallucinationERNIE 5.0 Thinking Preview by 3.9139.1+3.9
Images and chartsGemini 2.5 Pro
- MMMU-ProGemini 2.5 Pro by 10.364.674.9+10.3
Following instructionsGemini 2.5 Pro
- IFBenchGemini 2.5 Pro by 7.341.448.7+7.3
Other results8 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA Agentic IndexERNIE 5.0 Thinking Preview by 36.239.73.5+36.2
- τ²-Bench Telecom (AA run)ERNIE 5.0 Thinking Preview by 29.883.954.1+29.8
- AA-OmniscienceGemini 2.5 Pro by 29.3-45.6-16.4+29.3
- SimpleQAERNIE 5.0 Thinking Preview by 24.27550.8+24.2
- Artificial Analysis Coding IndexGemini 2.5 Pro by 17.529.246.7+17.5
- AIME 2025ERNIE 5.0 Thinking Preview by 48783+4
- MMLU-ProERNIE 5.0 Thinking Preview by 18786+1
- AA Intelligencetie14.315tie
Questions people ask
Which is better, ERNIE 5.0 Thinking Preview or Gemini 2.5 Pro?
Gemini 2.5 Pro wins five of the seven areas where both have results: coding, reasoning, images and charts, long documents and following instructions. ERNIE 5.0 Thinking Preview wins agents, and costs 60% less. They are level on facts.
Which is better for coding?
Gemini 2.5 Pro. It wins 2 of the 2 coding tests both models report; ERNIE 5.0 Thinking Preview wins none.
Which is cheaper?
ERNIE 5.0 Thinking Preview costs $0.89 per million input tokens and $3.58 per million output tokens; Gemini 2.5 Pro costs $1.25 and $10.00. That makes ERNIE 5.0 Thinking Preview about 60% cheaper for the same work.
How do you compare the two?
We use the 19 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 8 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.