---
title: "Cheaper alternatives to Gemini 2.5 Flash · Undominated.ai"
canonical: https://undominated.ai/alternatives/google__gemini-2.5-flash/
description: "28 models score at least as high as Gemini 2.5 Flash and cost no more on a balanced workload. 2 give up nothing measurable; the other 26 name what they drop. Measured prices and independent scores."
---

# Cheaper alternatives to Gemini 2.5 Flash · Undominated.ai

> 28 models score at least as high as Gemini 2.5 Flash and cost no more on a balanced workload. 2 give up nothing measurable; the other 26 name what they drop. Measured prices and independent scores.

# Cheaper alternatives to Gemini 2.5 Flash

Google · LMArena 1417.3 · $0.850/M on a balanced workload · prices as of 2026-08-28

## 2 models are both better and cheaper, giving up nothing.

Of the 133 models carrying both an independent score and a published price at the same delivery mode, **28** score at least as high as Gemini 2.5 Flash and cost no more under at least one workload. **27** of them are genuinely cheaper; the rest match the price and win on score alone. **2** match or beat it on every capability we hold — context, maximum output, input modes, tool use and extended reasoning. **26** do not, and every row names what it drops.

Dominance is not a property of a model. It is a property of a model, a capability lens and a workload, and all three are named on every row. Scores are LMArena Elo, used under CC BY 4.0 from the official dataset; prices are blended per million tokens. A higher score is not a drop-in replacement.

This page is built from that comparison and nothing else. When no model is both better and cheaper than Gemini 2.5 Flash, it is not generated.

## Where it holds

| Workload | Gemini 2.5 Flash $/M | Better and cheaper | With a trade |
| --- | --- | --- | --- |
| Balanced | $0.850 | 2 | 19 |
| Summarise | $0.333 | 1 | 14 |
| Chat | $1.10 | 2 | 24 |
| Code gen | $1.60 | 2 | 26 |
| Agentic | $0.469 | 2 | 16 |

Workload mixes are defined on the [methodology page](/methodology/). Cached input is priced at the cached rate, and reasoning tokens at the worse of the reasoning and output rates.

## What you would be replacing

| Intelligence | 1417.3 |
| --- | --- |
| Context window | 1.0M |
| Max output | 66K |
| Input modes | file, image, text, audio, video |
| Tool use | yes |
| Extended reasoning | yes |

A replacement has to clear every line above, not just the score. [Full record for Gemini 2.5 Flash](/models/google__gemini-2.5-flash/).

## Better and cheaper, nothing given up 2

Each of these matches or beats Gemini 2.5 Flash on context, maximum output, input modes, tool use and reasoning, scores at least as high, and costs no more.

| # | Model | Intelligence | $/M | Holds under |
| --- | --- | --- | --- | --- |
| 1 | Gemini 3.7 Flash Google | 1490.2 +72.9 | $0.750 | Balanced −12% Chat −20% Code gen −22% Agentic −15% |
| 2 | Gemini 3.5 Flash Lite Google | 1436.5 +19.2 | $0.850 | Balanced same price Summarise same price Chat same price Code gen same price Agentic same price |

## Cheaper and higher-scoring, but you give something up 26

These score at least as high and cost no more on the two plotted axes, and lose something that is not on them. Read the last column before switching.

| Model | Intelligence | $/M | Holds under | What you give up |
| --- | --- | --- | --- | --- |
| MiMo-V2.5-Pro Xiaomi | 1465 +47.7 | $0.544 | Balanced −36% Chat −56% Code gen −59% Agentic −48% | no file, image, audio, video input |
| Qwen3.7 Plus Qwen | 1456.2 +38.9 | $0.560 | Balanced −34% Summarise −11% Chat −43% Code gen −45% Agentic −34% | 1.0M → 1.0M context no file, audio, video input |
| Gemma 4 31B Google | 1441.7 +24.4 | $0.152 | Balanced −82% Summarise −73% Chat −84% Code gen −85% Agentic −78% | 1.0M → 262K context 66K → 16K max output no file, audio input |
| Hy3 Tencent | 1441.2 +23.9 | $0.144 | Balanced −83% Summarise −77% Chat −85% Code gen −86% Agentic −82% | 1.0M → 262K context no file, image, audio, video input |
| Gemma 4 26B A4B Google | 1434.6 +17.3 | $0.138 | Balanced −84% Summarise −75% Chat −84% Code gen −85% Agentic −76% | 1.0M → 262K context 66K → 16K max output no file, audio input |
| DeepSeek V4 Flash 0423 DeepSeek | 1431.6 +14.3 | $0.101 | Balanced −88% Summarise −80% Chat −91% Code gen −92% Agentic −88% | no file, image, audio, video input |
| MiniMax M3 MiniMax | 1434.8 +17.5 | $0.525 | Balanced −38% Summarise −17% Chat −46% Code gen −49% Agentic −38% | no file, audio input |
| Qwen3.6 Plus Qwen | 1436.8 +19.5 | $0.731 | Balanced −14% Chat −11% Code gen −19% | 1.0M → 1.0M context no file, audio input |
| GLM 4.7 Z.ai | 1435.3 +18 | $0.738 | Balanced −13% Chat −23% Code gen −26% Agentic −12% | 1.0M → 205K context no file, image, audio, video input |
| MiMo-V2.5 Xiaomi | 1427.3 tie | $0.175 | Balanced −79% Summarise −68% Chat −86% Code gen −87% Agentic −83% | no file input |
| GPT-5.6 Luna OpenAI | 1428.5 +11.2 | $0.450 | Balanced −47% Summarise −40% Chat −50% Code gen −51% Agentic −48% | no audio, video input |
| DeepSeek V3.2 DeepSeek | 1424.6 tie | $0.290 | Balanced −66% Summarise −31% Chat −76% Code gen −80% Agentic −57% | 1.0M → 164K context no file, image, audio, video input |
| DeepSeek V3.2 Exp DeepSeek | 1424.4 tie | $0.305 | Balanced −64% Summarise −17% Chat −70% Code gen −78% Agentic −38% | 1.0M → 164K context no file, image, audio, video input |
| Qwen3 235B A22B Instruct 2507 Qwen | 1419.3 tie | $0.205 | Balanced −76% Summarise −66% Chat −75% Code gen −77% Agentic −66% | 1.0M → 262K context 66K → 16K max output no file, image, audio, video input no extended reasoning |
| Qwen3 Next 80B A3B Instruct Qwen | 1418.6 tie | $0.350 | Balanced −59% Summarise −58% Chat −55% Code gen −56% Agentic −51% | 1.0M → 262K context no file, image, audio, video input no extended reasoning |
| DeepSeek V3.1 Terminus DeepSeek | 1419.6 tie | $0.453 | Balanced −47% Summarise −20% Chat −53% Code gen −56% Agentic −36% | 1.0M → 164K context 66K → 33K max output no file, image, audio, video input |
| Qwen3 VL 235B A22B Instruct Qwen | 1420.9 tie | $0.632 | Balanced −26% Summarise −21% Chat −22% Code gen −24% Agentic −15% | 1.0M → 262K context 66K → 33K max output no file, audio, video input no extended reasoning |
| Qwen3.5-122B-A10B Qwen | 1417.9 tie | $0.715 | Balanced −16% Chat −10% Code gen −15% | 1.0M → 262K context no file, audio input |
| DeepSeek V3.1 DeepSeek | 1419.1 tie | $0.825 | Balanced −3% Chat −10% Code gen −24% | 1.0M → 164K context no file, image, audio, video input |
| GLM 5 Z.ai | 1445.2 +27.9 | $0.984 chat | Chat −10% Code gen −15% | 1.0M → 205K context no file, image, audio, video input |
| Kimi K2.5 Moonshot AI | 1445.2 +27.9 | $1.06 chat | Chat −4% Code gen −6% | 1.0M → 262K context no file, audio, video input |
| GLM 4.6 Z.ai | 1439.8 +22.5 | $0.980 chat | Chat −11% Code gen −14% | 1.0M → 205K context no file, image, audio, video input |
| DeepSeek V4 Pro 0423 DeepSeek | 1439.2 +21.9 | $0.979 chat | Chat −11% Code gen −17% | no file, image, audio, video input |
| Qwen3.5 397B A17B Qwen | 1438.3 +21 | $1.56 code gen | Code gen −2% | 1.0M → 262K context no file, audio input |
| GLM 4.5 Z.ai | 1429.4 +12.1 | $1.09 chat | Chat −1% Code gen −5% | 1.0M → 131K context no file, image, audio, video input |
| R1 0528 DeepSeek | 1427.9 tie | $1.48 code gen | Code gen −8% | 1.0M → 164K context 66K → 33K max output no file, image, audio, video input |

## What this compares, and what it leaves out

 - Quality is LMArena Elo, used under CC BY 4.0 from the official dataset. The 95% confidence interval on a difference between two scores is about ±10.68 points, so a gap smaller than that is marked *tie* rather than an improvement — see [significance bands](/significance/).
 - 188 further models at this delivery mode carry a price but no independent score. They are absent from the comparison in both directions — unrated is not a zero, and an unmeasured model is neither an alternative nor a worse buy.
 - Retired models are never offered as an alternative, and a model only competes against its own delivery mode: batch trades latency for price, so it is not a like-for-like swap.
 - Nothing here measures latency, throughput, rate limits or how a model behaves on your prompts. Two models with the same index score are not interchangeable.
 - The models that nothing beats on both axes are on the [value frontier](/frontier/), and every other model something cheaper beats is [listed here](/alternatives/).