Pairwise benchmark snapshot · Aug 16, 2026noindex (thin pair)

DeepSeek: DeepSeek V3.2 vs Yi-Lightning benchmark

Name: DeepSeek: DeepSeek V3.2 vs Yi-Lightning AI benchmark comparison
Creator: CompareLLM
License: https://creativecommons.org/licenses/by/4.0/

This page compares DeepSeek: DeepSeek V3.2 (DeepSeek) and Yi-Lightning (01.AI) using the latest snapshots we have as of Aug 16, 2026. Yi-Lightning wins on output price at $0.14/1M. DeepSeek: DeepSeek V3.2 has the larger context window (164k tokens). Every cell below is a dated snapshot from a named public source — we do not invent missing scores, and we do not run SWE-bench ourselves.

Share Analysis:

WhatsApp Telegram X / Post LinkedIn Reddit

DeepSeek

DeepSeek: DeepSeek V3.2

Auto-discovered from OpenRouter (deepseek/deepseek-v3.2). Preview until a second source matches.

$0.39999999999999997/1M out

01.AI

Yi-Lightning

01.AI ultra-fast reasoning model delivering top LiveBench efficiency.

Elo 1,475$0.14/1M out

Swap or Add Any Model to this Matchup

Instant pairwise benchmark recalculation across 400+ LLMs

Model A

Model B

Quick vs:vs Claude Opus 4.5 vs Claude Opus 4.6 vs Claude Sonnet 4.5 vs Claude 3.7 Sonnet

Interactive Comparison Diagram

Matchup Breakdown

Tug-of-War Matchup Matrix

Category wins across intelligence, speed, and pricing efficiency.

DeepSeek: DeepSeek V3.2 (1)vsYi-Lightning (9)

DeepSeek: DeepSeek V3.2: 1W (10%)Overall: Yi-LightningYi-Lightning: 9W (90%)

← DeepSeek: DeepSeek V3.2Yi-Lightning →

Yi-Lightning (5/5)

Intelligence & Reasoning

Preference Elo, Coding proficiency, SWE-bench & LiveBench accuracy

Preference Elo

—vs1,475

Coding Elo

—vs1,480

SWE-bench

—vs63.4%

LiveBench

—vs65%

GPQA Diamond

—vs78%

DeepSeek: DeepSeek V3.2: 0WYi-Lightning: 5W

Yi-Lightning (2/2)

Speed & Latency

Generation throughput and time to first token responsiveness

Output speed

—vs130 tok/s

Time to first token

—vs150 ms

DeepSeek: DeepSeek V3.2: 0WYi-Lightning: 2W

Yi-Lightning (2/3)

Pricing & Capacity

Cost per million tokens and max context window length

Output price

$0.4/1Mvs$0.14/1M

Input price

$0.27/1Mvs$0.14/1M

Context window

164kvs16k

DeepSeek: DeepSeek V3.2: 1WYi-Lightning: 2W

Capability Matchup Radar

Multi-Dimensional Capability Radar

Each spoke is a skill. Farther from the center is better (0–100th percentile). Tap any dot to inspect details.

Tap any node to inspect

Percentile 0–100

Dimensional AdvantageYi-Lightning 5/6

Preference Elo

No shared data

Throughput Speed

No shared data

Price Efficiency

Yi-Lightning

$0.26/1M cheaper

Full Benchmark Score Matrix

—

Elo

High is better

—

EloHigh is better

1,475

Code Elo

High is better

—

Code EloHigh is better

1,480

LiveBench

High is better

—

LiveBenchHigh is better

65%

SWE-bench

High is better

—

SWE-benchHigh is better

63.4%

GPQA

High is better

—

GPQAHigh is better

78%

TTFT

Low is better

—

TTFTLow is better

150 ms

Speed

High is better

—

SpeedHigh is better

130 tok/s

In $

Low is better

$0.27/1M

In $Low is better

$0.14/1M+92%

Out $

Low is better

$0.4/1M

Out $Low is better

$0.14/1M+186%

Context

High is better

+900%164k

ContextHigh is better

16k

Interactive Simulator

Workload Cost & Savings Calculator

Simulate monthly production API costs based on published $/1M tokens.

Save up to 55% with Yi-Lightning

Monthly Volume50M tokens/mo

Quick Presets:

Token Ratio (In vs Out)70% In / 30% Out

RAG / Search (10% out)Coding / Chat (50% out)

DeepSeek: DeepSeek V3.2$15.42 / mo

In: $9.41Out: $6.00

Yi-Lightning$7.00 / mo

In: $4.90Out: $2.10

Estimated Cost Delta

Yi-Lightning is estimated to save $8.42/month ($101/year).

Based on 50M tokens (30% gen)

Key Selection Recommendations

1
Yi-Lightning:Pick Yi-Lightning when you are optimizing output cost.

Recommended Workload Routing

Repo / coding agentsinsufficient data

Need SWE-bench on both sides.

High-volume chatYi-Lightning

Lower output list price ($0.14/1M).

Voice / low-latency UIinsufficient data

Need TTFT on both sides.

Long-document RAGDeepSeek: DeepSeek V3.2

Larger window (164k).

Screenshots / visionYi-Lightning

Yi-Lightning is the side marked multimodal in the catalog.

Target Workload	Recommended Pick	Evaluation Rationale
Repo / coding agents	insufficient data	Need SWE-bench on both sides.
High-volume chat	Yi-Lightning	Lower output list price ($0.14/1M).
Voice / low-latency UI	insufficient data	Need TTFT on both sides.
Long-document RAG	DeepSeek: DeepSeek V3.2	Larger window (164k).
Screenshots / vision	Yi-Lightning	Yi-Lightning is the side marked multimodal in the catalog.

Community Sentiment

Cast Your Matchup Vote

0 total votes

Community Discussions (0)

No comments posted on this matchup yet. Be the first to share an evaluation note!

Share card

The real PNG is generated at /compare/deepseek-v3-2-vs-yi-lightning/opengraph-image for crawlers.

PNG

CompareLLM Matrix

Verified head-to-head card

LIVE

DeepSeek

DeepSeek: DeepSeek V3.2

Elo n/a

LiveBench—

SWE-bench—

Speed—

Output cost$0.4/1M

01.AI

Yi-Lightning

Elo 1,475

LiveBench65%

SWE-bench63.4%

Speed130 tok/s

Output cost$0.14/1M

Frequently Asked Questions

Which is better overall, DeepSeek: DeepSeek V3.2 or Yi-Lightning?: We do not have preference Elo for both models, so we do not declare an overall winner. Compare the metrics that exist.
Which is better at coding, DeepSeek: DeepSeek V3.2 or Yi-Lightning?: SWE-bench is missing for at least one model, so we do not rank coding from a single number.
Which is cheaper to run in production?: Yi-Lightning output tokens are $0.14/1M versus $0.4/1M. Input prices and retry rates still move the real bill.
Which is faster, DeepSeek: DeepSeek V3.2 or Yi-Lightning?: We do not have TTFT for both models.
Which has the larger context window?: DeepSeek: DeepSeek V3.2 accepts 164k tokens versus 16k.
Can I self-host either model?: Neither model is marked open-weights here. You are comparing hosted APIs.
Are these scores from CompareLLM’s own evals?: No. V1 aggregates public leaderboards (and optional first-party latency pings). Each cell names its source and date. Read /methodology.
What does preference Elo mean on this page?: Preference Elo is a crowd vote from LMArena / Arena. People see two hidden answers and pick the one they like more. The model that wins more often gets a higher Elo. That means people preferred it — not that it passed a school test. It is not SWE-bench, not accuracy, and not a number we invent. DeepSeek: DeepSeek V3.2 and Yi-Lightning Elo cells are dated snapshots from the named source — we do not compute Arena ourselves.

Similar Head-to-Head Comparisons

All DeepSeek: DeepSeek V3.2 matchups →|All Yi-Lightning matchups →

Yi-Lightning vs Yi-Large (previous other)Yi-Lightning vs Ernie 4.5 Turbo (next other)DeepSeek V4 Pro vs Yi-Lightning DeepSeek V4 Flash vs Yi-Lightning Claude Opus 4.5 vs Yi-Lightning Claude Opus 4.6 vs Yi-Lightning

DeepSeek: DeepSeek V3.2 vs Yi-Lightning benchmark

Interactive Comparison Diagram

Matchup Breakdown

Tug-of-War Matchup Matrix

Category wins across intelligence, speed, and pricing efficiency.

DeepSeek: DeepSeek V3.2 (1)vsYi-Lightning (9)

DeepSeek: DeepSeek V3.2: 1W (10%)Overall: Yi-LightningYi-Lightning: 9W (90%)

← DeepSeek: DeepSeek V3.2Yi-Lightning →

Yi-Lightning (5/5)

Intelligence & Reasoning

Preference Elo, Coding proficiency, SWE-bench & LiveBench accuracy

Preference Elo

—vs1,475

Coding Elo

—vs1,480

SWE-bench

—vs63.4%

LiveBench

—vs65%

GPQA Diamond

—vs78%

DeepSeek: DeepSeek V3.2: 0WYi-Lightning: 5W

Yi-Lightning (2/2)

Speed & Latency

Generation throughput and time to first token responsiveness

Output speed

—vs130 tok/s

Time to first token

—vs150 ms

DeepSeek: DeepSeek V3.2: 0WYi-Lightning: 2W

Yi-Lightning (2/3)

Pricing & Capacity

Cost per million tokens and max context window length

Output price

$0.4/1Mvs$0.14/1M

Input price

$0.27/1Mvs$0.14/1M

Context window

164kvs16k

DeepSeek: DeepSeek V3.2: 1WYi-Lightning: 2W

Recommended Workload Routing

Repo / coding agentsinsufficient data

Need SWE-bench on both sides.

High-volume chatYi-Lightning

Lower output list price ($0.14/1M).

Voice / low-latency UIinsufficient data

Need TTFT on both sides.

Long-document RAGDeepSeek: DeepSeek V3.2

Larger window (164k).

Screenshots / visionYi-Lightning

Yi-Lightning is the side marked multimodal in the catalog.

Target Workload	Recommended Pick	Evaluation Rationale
Repo / coding agents	insufficient data	Need SWE-bench on both sides.
High-volume chat	Yi-Lightning	Lower output list price ($0.14/1M).
Voice / low-latency UI	insufficient data	Need TTFT on both sides.
Long-document RAG	DeepSeek: DeepSeek V3.2	Larger window (164k).
Screenshots / vision	Yi-Lightning	Yi-Lightning is the side marked multimodal in the catalog.

Frequently Asked Questions

Which is better overall, DeepSeek: DeepSeek V3.2 or Yi-Lightning?

We do not have preference Elo for both models, so we do not declare an overall winner. Compare the metrics that exist.

Which is better at coding, DeepSeek: DeepSeek V3.2 or Yi-Lightning?

SWE-bench is missing for at least one model, so we do not rank coding from a single number.

Which is cheaper to run in production?

Yi-Lightning output tokens are $0.14/1M versus $0.4/1M. Input prices and retry rates still move the real bill.

Which is faster, DeepSeek: DeepSeek V3.2 or Yi-Lightning?

We do not have TTFT for both models.

Which has the larger context window?

DeepSeek: DeepSeek V3.2 accepts 164k tokens versus 16k.

Can I self-host either model?

Neither model is marked open-weights here. You are comparing hosted APIs.

Are these scores from CompareLLM’s own evals?

No. V1 aggregates public leaderboards (and optional first-party latency pings). Each cell names its source and date. Read /methodology.

What does preference Elo mean on this page?

Preference Elo is a crowd vote from LMArena / Arena. People see two hidden answers and pick the one they like more. The model that wins more often gets a higher Elo. That means people preferred it — not that it passed a school test. It is not SWE-bench, not accuracy, and not a number we invent. DeepSeek: DeepSeek V3.2 and Yi-Lightning Elo cells are dated snapshots from the named source — we do not compute Arena ourselves.