---
title: "Cheaper alternatives to Llama 3.1 70B Instruct · Undominated.ai"
canonical: https://undominated.ai/alternatives/meta-llama__llama-3.1-70b-instruct/
description: "56 models score at least as high as Llama 3.1 70B Instruct and cost no more on a balanced workload. 50 give up nothing measurable; the other 6 name what they drop. Measured prices and independent scores."
---

# Cheaper alternatives to Llama 3.1 70B Instruct · Undominated.ai

> 56 models score at least as high as Llama 3.1 70B Instruct and cost no more on a balanced workload. 50 give up nothing measurable; the other 6 name what they drop. Measured prices and independent scores.

# Cheaper alternatives to Llama 3.1 70B Instruct

Meta · LMArena 1261 · $0.400/M on a balanced workload · prices as of 2026-08-28

## 50 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, **56** score at least as high as Llama 3.1 70B Instruct and cost no more under at least one workload. **50** match or beat it on every capability we hold — context, maximum output, input modes, tool use and extended reasoning. **6** 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 Llama 3.1 70B Instruct, it is not generated.

## Where it holds

| Workload | Llama 3.1 70B Instruct $/M | Better and cheaper | With a trade |
| --- | --- | --- | --- |
| Balanced | $0.400 | 25 | 5 |
| Summarise | $0.400 | 50 | 6 |
| Chat | $0.400 | 23 | 5 |
| Code gen | $0.400 | 22 | 3 |
| Agentic | $0.400 | 44 | 6 |

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 | 1261 |
| --- | --- |
| Context window | 131K |
| Max output | 16K |
| Input modes | text |
| Tool use | yes |
| Extended reasoning | no |

A replacement has to clear every line above, not just the score. [Full record for Llama 3.1 70B Instruct](/models/meta-llama__llama-3.1-70b-instruct/).

## Better and cheaper, nothing given up 50

Each of these matches or beats Llama 3.1 70B Instruct 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 | Gemma 4 31B Google | 1441.7 +180.7 | $0.152 | Balanced −62% Summarise −77% Chat −56% Code gen −41% Agentic −74% |
| 2 | Hy3 Tencent | 1441.2 +180.2 | $0.144 | Balanced −64% Summarise −81% Chat −59% Code gen −43% Agentic −79% |
| 3 | Gemma 4 26B A4B Google | 1434.6 +173.6 | $0.138 | Balanced −66% Summarise −79% Chat −56% Code gen −42% Agentic −72% |
| 4 | DeepSeek V4 Flash 0423 DeepSeek | 1431.6 +170.6 | $0.101 | Balanced −75% Summarise −83% Chat −77% Code gen −69% Agentic −86% |
| 5 | MiMo-V2.5 Xiaomi | 1427.3 +166.3 | $0.175 | Balanced −56% Summarise −73% Chat −61% Code gen −47% Agentic −80% |
| 6 | DeepSeek V3.2 DeepSeek | 1424.6 +163.6 | $0.290 | Balanced −28% Summarise −43% Chat −33% Code gen −20% Agentic −50% |
| 7 | DeepSeek V3.2 Exp DeepSeek | 1424.4 +163.4 | $0.305 | Balanced −24% Summarise −31% Chat −19% Code gen −12% Agentic −27% |
| 8 | Qwen3 235B A22B Instruct 2507 Qwen | 1419.3 +158.3 | $0.205 | Balanced −49% Summarise −72% Chat −32% Code gen −9% Agentic −60% |
| 9 | Qwen3 Next 80B A3B Instruct Qwen | 1418.6 +157.6 | $0.350 | Balanced −13% Summarise −65% Agentic −42% |
| 10 | Step 3.5 Flash StepFun | 1403.8 +142.8 | $0.150 | Balanced −63% Summarise −73% Chat −55% Code gen −45% Agentic −68% |
| 11 | Qwen3.5-Flash Qwen | 1397.6 +136.6 | $0.114 | Balanced −72% Summarise −81% Chat −64% Code gen −55% Agentic −76% |
| 12 | Qwen3 30B A3B Instruct 2507 Qwen | 1384.3 +123.3 | $0.084 | Balanced −79% Summarise −86% Chat −73% Code gen −66% Agentic −83% |
| 13 | GLM 4.5 Air Z.ai | 1382.8 +121.8 | $0.310 | Balanced −23% Summarise −66% Chat −3% Agentic −56% |
| 14 | Solar Pro 4 Upstage | 1376.2 +115.2 | $0.052 | Balanced −87% Summarise −93% Chat −85% Code gen −79% Agentic −93% |
| 15 | gpt-oss-120b OpenAI | 1365.6 +104.6 | $0.070 | Balanced −82% Summarise −89% Chat −77% Code gen −71% Agentic −86% |
| 16 | Gemma 3 27B Google | 1358.3 +97.3 | $0.172 | Balanced −57% Summarise −78% Chat −46% Code gen −25% Agentic −72% |
| 17 | GLM 4.7 Flash Z.ai | 1352.9 +91.9 | $0.145 | Balanced −64% Summarise −84% Chat −55% Code gen −35% Agentic −80% |
| 18 | Qwen3 32B Qwen | 1340.1 +79.1 | $0.130 | Balanced −68% Summarise −78% Chat −60% Code gen −50% Agentic −73% |
| 19 | Trinity Large Thinking Arcee | 1341.9 +80.9 | $0.378 | Balanced −6% Summarise −49% Agentic −45% |
| 20 | Gemma 3 12B Google | 1334.2 +73.2 | $0.075 | Balanced −81% Summarise −86% Chat −78% Code gen −73% Agentic −84% |
| 21 | GPT-5 Nano OpenAI | 1320.3 +59.3 | $0.138 | Balanced −66% Summarise −86% Chat −56% Code gen −36% Agentic −81% |
| 22 | Qwen3 30B A3B Qwen | 1316.9 +55.9 | $0.215 | Balanced −46% Summarise −65% Chat −32% Code gen −13% Agentic −56% |
| 23 | Granite 4.1 8B IBM | 1291.6 +30.6 | $0.063 | Balanced −84% Summarise −87% Chat −83% Code gen −80% Agentic −86% |
| 24 | gpt-oss-20b OpenAI | 1287.8 +26.8 | $0.055 | Balanced −86% Summarise −91% Chat −83% Code gen −78% Agentic −89% |
| 25 | GPT-4.1 Nano OpenAI | 1284.8 +23.8 | $0.175 | Balanced −56% Summarise −77% Chat −51% Code gen −32% Agentic −75% |
| 26 | Gemini 3.7 Flash Google | 1490.2 +229.2 | $0.354 summarise | Summarise −12% Agentic under 1% less |
| 27 | MiMo-V2.5-Pro Xiaomi | 1465 +204 | $0.334 summarise | Summarise −17% Agentic −39% |
| 28 | Qwen3.7 Plus Qwen | 1456.2 +195.2 | $0.295 summarise | Summarise −26% Agentic −22% |
| 29 | Gemini 3.5 Flash Lite Google | 1436.5 +175.5 | $0.333 summarise | Summarise −17% |
| 30 | MiniMax M3 MiniMax | 1434.8 +173.8 | $0.277 summarise | Summarise −31% Agentic −27% |
| 31 | GLM 4.7 Z.ai | 1435.3 +174.3 | $0.376 summarise | Summarise −6% |
| 32 | GPT-5.6 Luna OpenAI | 1428.5 +167.5 | $0.199 summarise | Summarise −50% Agentic −39% |
| 33 | Qwen3 VL 235B A22B Instruct Qwen | 1420.9 +159.9 | $0.263 summarise | Summarise −34% Agentic −1% |
| 34 | DeepSeek V3.1 Terminus DeepSeek | 1419.6 +158.6 | $0.268 summarise | Summarise −33% Agentic −25% |
| 35 | Qwen3.5-122B-A10B Qwen | 1417.9 +156.9 | $0.351 summarise | Summarise −12% |
| 36 | Gemini 2.5 Flash Google | 1417.3 +156.3 | $0.333 summarise | Summarise −17% |
| 37 | Gemini 3.1 Flash Lite Preview Google | 1414.8 +153.8 | $0.248 summarise | Summarise −38% Agentic −24% |
| 38 | Qwen3 235B A22B Thinking 2507 Qwen | 1413.8 +152.8 | $0.334 summarise | Summarise −17% |
| 39 | Inkling Small Thinking Machines | 1411.7 +150.7 | $0.388 summarise | Summarise −3% Agentic −11% |
| 40 | Qwen3.5-27B Qwen | 1407.9 +146.9 | $0.263 summarise | Summarise −34% Agentic under 1% less |
| 41 | MiniMax M2.7 MiniMax | 1405.3 +144.3 | $0.277 summarise | Summarise −31% Agentic −27% |
| 42 | Qwen3.5-35B-A3B Qwen | 1395.6 +134.6 | $0.300 summarise | Summarise −25% Agentic same price |
| 43 | GLM 4.6V Z.ai | 1374.7 +113.7 | $0.260 summarise | Summarise −35% Agentic −39% |
| 44 | DeepSeek V3 0324 DeepSeek | 1375 +114 | $0.287 summarise | Summarise −28% Agentic −9% |
| 45 | GPT-5.4 Nano OpenAI | 1372.8 +111.8 | $0.201 summarise | Summarise −50% Agentic −37% |
| 46 | GPT-5 Mini OpenAI | 1373.4 +112.4 | $0.273 summarise | Summarise −32% Agentic −5% |
| 47 | Qwen3 Next 80B A3B Thinking Qwen | 1367.5 +106.5 | $0.203 summarise | Summarise −49% Agentic −23% |
| 48 | MiniMax M2.5 MiniMax | 1359 +98 | $0.241 summarise | Summarise −40% Agentic −38% |
| 49 | MiniMax M2 MiniMax | 1342.1 +81.1 | $0.293 summarise | Summarise −27% Agentic −8% |
| 50 | GPT-4.1 Mini OpenAI | 1340.5 +79.5 | $0.374 summarise | Summarise −6% |

## Cheaper and higher-scoring, but you give something up 6

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 |
| --- | --- | --- | --- | --- |
| Mercury 2 Inception | 1357.8 +96.8 | $0.375 | Balanced −6% Summarise −47% Chat −4% Agentic −52% | 131K → 128K context |
| Olmo 3 32B Think Allen AI | 1298.5 +37.5 | $0.237 | Balanced −41% Summarise −58% Chat −28% Code gen −10% Agentic −49% | 131K → 66K context no tool use |
| Gemma 3 4B Google | 1290.8 +29.8 | $0.063 | Balanced −84% Summarise −87% Chat −83% Code gen −80% Agentic −86% | no tool use |
| GPT-4o-mini (2024-07-18) OpenAI | 1286.6 +25.6 | $0.263 | Balanced −34% Summarise −62% Chat −23% Agentic −57% | 131K → 128K context |
| Qwen2.5 72B Instruct Qwen | 1269.1 tie | $0.370 | Balanced −8% Summarise −10% Chat −6% Code gen −4% Agentic −9% | 131K → 33K context |
| DeepSeek V3 DeepSeek | 1332.6 +71.6 | $0.296 summarise | Summarise −26% Agentic −7% | 16K → 16K max output |

## 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/).