Is Gemma 2 27B a good deal?

Whether anything in this catalogue beats Gemma 2 27B on both quality and price, and what you give up if it does. A computation on the current catalogue, not an opinion.

13 undominated of 136 · Sep 23, 2026

As of Sep 23, 2026, Gemma 2 27B is dominated for Balanced on LMArena. GLM 5.3 Flash scores 240.6 higher and costs 63% less, with a covering envelope. 13 of 136 rated, priced standard models are undominated.

Inspect model evidence Compare differences & requirements

Current model

Gemma 2 27B

Balanced

3 tokens in per 1 out

GLM 5.3 Flash is both better and cheaper than Gemma 2 27B: 240.6 points higher on LMArena and 63% less per million tokens, $0.41 cheaper at this mix.

Gemma 2 27B

LMArena Elo · higher is better

Scale starts at 1040 Elo

1231.3

Effective $/M · Balanced · lower is better

$0.650/M

GLM 5.3 Flash

LMArena Elo · higher is better

Scale starts at 1040 Elo

1471.9

Effective $/M · Balanced · lower is better

$0.237/M

Gemma 2 27B takes text, returns up to 2,048 tokens from a 8,192-token context, and is listed by 1 seller.

Compared against 136 rated, priced models on this lens: 52 models dominate it and give up nothing, none does so with a trade-off. Gemma 2 27B scores 1231.3 at $0.65 per million tokens for this mix.

Envelope-safe replacements

Each row scores at least as high, costs no more, and covers this model’s context, output, modalities, tools, and reasoning. A cheaper narrower model is not listed here.

ModelLMArenaEffective $/MYou save
GLM 5.3 Flash1471.9 +240.6$0.237/M63%
MiMo-V2.5-Pro1464.8 +233.5$0.544/M16%
Qwen3.7 Plus1454.2 +222.9$0.560/M14%
Gemma 4 31B1441.7 +210.4$0.153/M77%
Hy31440.6 +209.3$0.144/M78%
Gemma 4 26B A4B 1434.5 +203.2$0.143/M78%
MiniMax M31433.5 +202.2$0.525/M19%
DeepSeek V4 Flash 04231431.8 +200.5$0.103/M84%
GPT-5.6 Luna1429.9 +198.6$0.450/M31%
MiMo-V2.51427.4 +196.1$0.175/M73%
DeepSeek V3.21424.8 +193.5$0.302/M54%
DeepSeek V3.2 Exp1422.6 +191.3$0.305/M53%
Qwen3 VL 235B A22B Instruct1420.6 +189.3$0.632/M3%
Qwen3 235B A22B Instruct 25071419.8 +188.5$0.153/M76%
DeepSeek V3.11419.7 +188.4$0.425/M35%
Qwen3 Next 80B A3B Instruct1417.6 +186.3$0.343/M47%
DeepSeek V3.1 Terminus1416.9 +185.6$0.453/M30%
Gemini 3.1 Flash Lite Preview1415.3 +184.0$0.563/M13%
Inkling Small1412.3 +181.0$0.637/M2%
Qwen3.5-27B1408.1 +176.8$0.536/M18%
MiniMax M2.71404.6 +173.3$0.525/M19%
Step 3.5 Flash1403.7 +172.4$0.150/M77%
Qwen3.5-Flash1397.7 +166.4$0.114/M83%
Qwen3.5-35B-A3B1395.5 +164.2$0.547/M16%
GLM 4.5 Air1383.7 +152.4$0.310/M52%
Qwen3 30B A3B Instruct 25071383.6 +152.3$0.084/M87%
Solar Pro 41377.3 +146.0$0.158/M76%
GLM 4.6V1376.5 +145.2$0.450/M31%
DeepSeek V3 03241375.0 +143.7$0.438/M33%
GPT-5.4 Nano1372.9 +141.6$0.463/M29%
Qwen3 Next 80B A3B Thinking1367.8 +136.5$0.412/M37%
gpt-oss-120b1365.9 +134.6$0.262/M60%
MiniMax M2.51358.9 +127.6$0.473/M27%
Gemma 3 27B1357.7 +126.4$0.172/M73%
Mercury 21357.6 +126.3$0.375/M42%
GLM 4.7 Flash1352.2 +120.9$0.145/M78%
MiniMax M21342.1 +110.8$0.446/M31%
Trinity Large Thinking1341.8 +110.5$0.388/M40%
Qwen3 32B1340.0 +108.7$0.130/M80%
Gemma 3 12B1334.2 +102.9$0.075/M88%
DeepSeek V31332.4 +101.1$0.463/M29%
GPT-5 Nano1319.7 +88.4$0.138/M79%
Qwen3 30B A3B1317.0 +85.7$0.215/M67%
Granite 4.2 8B1316.8 +85.5$0.107/M83%
Gemma 3 4B1290.7 +59.4$0.063/M90%
gpt-oss-20b1287.3 +56.0$0.036/M94%
GPT-4o-mini (2024-07-18)1286.3 +55.0$0.262/M60%
GPT-4.1 Nano1284.7 +53.4$0.175/M73%
Llama 3.3 70B Instruct1274.0 +42.7$0.155/M76%
Qwen2.5 72B Instruct1268.9 +37.6$0.370/M43%
Llama 3.1 70B Instruct1260.8 +29.5$0.400/M38%
Mistral Small 31233.5 +2.2$0.058/M91%
GLM 5.3 Flash1471.9 · $0.237/MGemma 2 27B1231.3 · $0.650/MLMArena EloEffective $/M · Balanced
136 rated, priced standard models. Chartreuse is the frontier. Cobalt is the model you named, when it is not on the staircase.

Open the frontier

The link keeps the model and mix, using the current catalogue. Monthly spend and switching cost stay in this browser tab and are left out of the link.

Saved decision references

Save the model, workload, catalogue date and capability-preservation rule in this browser. Spend, switching cost and bill contents are not saved. No account or notifications.

Constraint: replacements must preserve the model’s capabilities; any losses remain named trade-offs.

Historical decisions cannot be fully replayed from saved references: past prices, scores and capability evidence are not stored.

Evidence & Ask