Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d271Policy30d66, 14 pendingSources1,015

ERNIE 5.0 Thinking Preview vs Qwen3 VL 235B A22B Instruct

Baidu · released

Wins 4 of 7 areas

Coding · Agents · Reasoning · Facts

Alibaba · released

Wins 3 of 7 areas

Images and charts · Long documents · Following instructions

ERNIE 5.0 Thinking Preview wins more areas, narrowly.Qwen3 VL 235B A22B Instruct is cheaper and better at images and charts.

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.

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$0.89 to read · $3.58 to write$4.47Price from Baidu's own price page
Qwen3 VL 235B A22B Instruct$0.40 to read · $1.60 to write$2.00Price from Alibaba's own price page

Qwen3 VL 235B A22B Instruct costs 55% less for the same work.

The biggest differences

The three tests each model wins by the widest margin. Scores are out of 100.

Where ERNIE 5.0 Thinking Preview pulls ahead

  • Hard command-line tasks in a real terminalTerminal-Bench Hard+18.2points aheadERNIE 5.0 Thinking Preview25Qwen3 VL 235B A22B Instruct6.8
  • Real work tasks from 44 professionsGDPVal+7.7points aheadERNIE 5.0 Thinking Preview14.4Qwen3 VL 235B A22B Instruct6.7
  • Very hard expert questions across many subjectsHumanity's Last Exam+6.7points aheadERNIE 5.0 Thinking Preview13.3Qwen3 VL 235B A22B Instruct6.6

Where Qwen3 VL 235B A22B Instruct pulls ahead

  • Reasons across sets of long documentsAA-LCR+25.7points aheadQwen3 VL 235B A22B Instruct32.7ERNIE 5.0 Thinking Preview7
  • Harder college exam questions with imagesMMMU-Pro+3points aheadQwen3 VL 235B A22B Instruct67.6ERNIE 5.0 Thinking Preview64.6
  • Follows unfamiliar, precisely checkable instructionsIFBench+1.3points aheadQwen3 VL 235B A22B Instruct42.7ERNIE 5.0 Thinking Preview41.4

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.

CodingERNIE 5.0 Thinking Preview

Full coding ranking

AgentsERNIE 5.0 Thinking Preview
  • GDPValERNIE 5.0 Thinking Preview by 7.714.46.7+7.7

Full agents ranking

ReasoningERNIE 5.0 Thinking Preview

Full reasoning ranking

FactsERNIE 5.0 Thinking Preview

Full facts ranking

Images and chartsQwen3 VL 235B A22B Instruct
  • MMMU-ProQwen3 VL 235B A22B Instruct by 364.667.6+3

Full images and charts ranking

Long documentsQwen3 VL 235B A22B Instruct
  • AA-LCRQwen3 VL 235B A22B Instruct by 25.7732.7+25.7

Full long documents ranking

Following instructionsQwen3 VL 235B A22B Instruct
  • IFBenchQwen3 VL 235B A22B Instruct by 1.341.442.7+1.3

Full following instructions ranking

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.

Questions people ask

Which is better, ERNIE 5.0 Thinking Preview or Qwen3 VL 235B A22B Instruct?

ERNIE 5.0 Thinking Preview wins four of the seven areas where both have results: coding, agents, reasoning and facts. Qwen3 VL 235B A22B Instruct wins images and charts, long documents and following instructions, and costs 55% less.

Which is better for coding?

ERNIE 5.0 Thinking Preview. It wins 2 of the 2 coding tests both models report; Qwen3 VL 235B A22B Instruct wins none.

Which is cheaper?

ERNIE 5.0 Thinking Preview costs $0.89 per million input tokens and $3.58 per million output tokens; Qwen3 VL 235B A22B Instruct costs $0.40 and $1.60. That makes Qwen3 VL 235B A22B Instruct about 55% 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.