The uncomfortable truth: “model choice” is half your prompt engineering If your prompt is a recipe, the model is your kitchen. A great recipe doesn’t help if: the oven is tiny (context window),
the ingredients are expensive (token price),
the chef is slow (latency),
or your tools don’t fit (function calling / JSON / SDK / ecosystem). the oven is tiny (context window), the ingredients are expensive (token price), the chef is slow (latency), or your tools don’t fit (function calling / JSON / SDK / ecosystem). So here’s a practical comparison you can actually use. practical Note on “parameters”: for many frontier models, parameter counts are not publicly disclosed. In practice, context window + pricing + tool features predict “fit” better than guessing parameter scale. Note on “parameters”: for many frontier models, parameter counts are not publicly disclosed. In practice, context window + pricing + tool features predict “fit” better than guessing parameter scale. Note on “parameters”: 1) Quick comparison: what you should care about first 1.1 The “four knobs” that matter Context: can you fit the job in one request?
Cost: can you afford volume?
Latency: does your UX tolerate the wait?
Compatibility: will your stack integrate cleanly? Context: can you fit the job in one request? Context Cost: can you afford volume? Cost Latency: does your UX tolerate the wait? Latency Compatibility: will your stack integrate cleanly? Compatibility Everything else is second order. 2) Model spec table (context + positioning) This table focuses on what’s stable: family, positioning, and context expectations. family, positioning, and context expectations | Provider | Model family (examples) | Typical positioning | Notes | |---|---|---| | OpenAI | GPT family (e.g., gpt-4o, gpt-4.1, gpt-5*) | General-purpose, strong tooling ecosystem | Pricing + cached input are clearly published.  | OpenAI | “o” reasoning family (e.g., o3, o1) | Deep reasoning / harder planning | Often higher cost; use selectively.  | Anthropic | Claude family (e.g., Haiku / Sonnet tiers) | Strong writing + safety posture; clean docs | Pricing table includes multiple rate dimensions.  | Google | Gemini family (Flash / Pro tiers) | Multimodal + Google ecosystem + caching/grounding options | Pricing page explicitly covers caching + grounding.  | DeepSeek | DeepSeek chat + reasoning models | Aggressive price/perf, popular for scale | Official pricing docs available.  | Open source | Llama / Qwen / Mistral etc. | Self-host for privacy/control | Context depends on model; Llama 3.1 supports 128K. gpt-4o gpt-4.1 gpt-5* o3 o1 3) Pricing table (the part your CFO actually reads) Below are public list prices from official docs (USD per 1M tokens).   Use this as a baseline, then apply: caching, batch discounts, and your real output length. public list prices 1M tokens 3.1 OpenAI (selected highlights) OpenAI publishes input, cached input, and output prices per 1M tokens. Model

Input / 1M

Cached input / 1M

Output / 1M

When to use



gpt-4.1

$2.00

$0.50

$8.00

High-quality general reasoning with sane cost



gpt-4o

$2.50

$1.25

$10.00

Multimodal-ish “workhorse” if you need it



gpt-4o-mini

$0.15

$0.075

$0.60

High-throughput chat, extraction, tagging



o3

$2.00

$0.50

$8.00

Reasoning-heavy tasks without the top-end pricing



o1

$15.00

$7.50

$60.00

“Use sparingly”: hard reasoning where mistakes are expensive Model

Input / 1M

Cached input / 1M

Output / 1M

When to use



gpt-4.1

$2.00

$0.50

$8.00

High-quality general reasoning with sane cost



gpt-4o

$2.50

$1.25

$10.00

Multimodal-ish “workhorse” if you need it



gpt-4o-mini

$0.15

$0.075

$0.60

High-throughput chat, extraction, tagging



o3

$2.00

$0.50

$8.00

Reasoning-heavy tasks without the top-end pricing



o1

$15.00

$7.50

$60.00

“Use sparingly”: hard reasoning where mistakes are expensive Model

Input / 1M

Cached input / 1M

Output / 1M

When to use Model Model Input / 1M Input / 1M Cached input / 1M Cached input / 1M Output / 1M Output / 1M When to use When to use gpt-4.1

$2.00

$0.50

$8.00

High-quality general reasoning with sane cost gpt-4.1 gpt-4.1 gpt-4.1 $2.00 $2.00 $0.50 $0.50 $8.00 $8.00 High-quality general reasoning with sane cost High-quality general reasoning with sane cost gpt-4o

$2.50

$1.25

$10.00

Multimodal-ish “workhorse” if you need it gpt-4o gpt-4o gpt-4o $2.50 $2.50 $1.25 $1.25 $10.00 $10.00 Multimodal-ish “workhorse” if you need it Multimodal-ish “workhorse” if you need it gpt-4o-mini

$0.15

$0.075

$0.60

High-throughput chat, extraction, tagging gpt-4o-mini gpt-4o-mini gpt-4o-mini $0.15 $0.15 $0.075 $0.075 $0.60 $0.60 High-throughput chat, extraction, tagging High-throughput chat, extraction, tagging o3

$2.00

$0.50

$8.00

Reasoning-heavy tasks without the top-end pricing o3 o3 o3 $2.00 $2.00 $0.50 $0.50 $8.00 $8.00 Reasoning-heavy tasks without the top-end pricing Reasoning-heavy tasks without the top-end pricing o1

$15.00

$7.50

$60.00

“Use sparingly”: hard reasoning where mistakes are expensive o1 o1 o1 $15.00 $15.00 $7.50 $7.50 $60.00 $60.00 “Use sparingly”: hard reasoning where mistakes are expensive “Use sparingly”: hard reasoning where mistakes are expensive If you’re building a product: you’ll often run 80–95% of calls on a cheaper model (mini/fast tier), and escalate only the hard cases. If you’re building a product: you’ll often run 80–95% of calls on a cheaper model (mini/fast tier), and escalate only the hard cases. 80–95% 3.2 Anthropic (Claude) Anthropic publishes a model pricing table in Claude docs. Model

Input / MTok

Output / MTok

Notes



Claude Haiku 4.5

$1.00

$5.00

Fast, budget-friendly tier



Claude Haiku 3.5

$0.80

$4.00

Even cheaper tier option



Claude Sonnet 3.7 (deprecated)

$3.75

$15.00

Listed as deprecated on pricing



Claude Opus 3 (deprecated)

$18.75

$75.00

Premium, but marked deprecated Model

Input / MTok

Output / MTok

Notes



Claude Haiku 4.5

$1.00

$5.00

Fast, budget-friendly tier



Claude Haiku 3.5

$0.80

$4.00

Even cheaper tier option



Claude Sonnet 3.7 (deprecated)

$3.75

$15.00

Listed as deprecated on pricing



Claude Opus 3 (deprecated)

$18.75

$75.00

Premium, but marked deprecated Model

Input / MTok

Output / MTok

Notes Model Model Input / MTok Input / MTok Output / MTok Output / MTok Notes Notes Claude Haiku 4.5

$1.00

$5.00

Fast, budget-friendly tier Claude Haiku 4.5 Claude Haiku 4.5 $1.00 $1.00 $5.00 $5.00 Fast, budget-friendly tier Fast, budget-friendly tier Claude Haiku 3.5

$0.80

$4.00

Even cheaper tier option Claude Haiku 3.5 Claude Haiku 3.5 $0.80 $0.80 $4.00 $4.00 Even cheaper tier option Even cheaper tier option Claude Sonnet 3.7 (deprecated)

$3.75

$15.00

Listed as deprecated on pricing Claude Sonnet 3.7 (deprecated) Claude Sonnet 3.7 (deprecated) $3.75 $3.75 $15.00 $15.00 Listed as deprecated on pricing Listed as deprecated on pricing Claude Opus 3 (deprecated)

$18.75

$75.00

Premium, but marked deprecated Claude Opus 3 (deprecated) Claude Opus 3 (deprecated) $18.75 $18.75 $75.00 $75.00 Premium, but marked deprecated Premium, but marked deprecated Important: model availability changes. Treat the pricing table as the authoritative “what exists right now.” Important: model availability changes. Treat the pricing table as the authoritative “what exists right now.” Important: 3.3 Google Gemini (Developer API) Gemini pricing varies by tier and includes context caching + grounding pricing. Tier (example rows from pricing page)

Input / 1M (text/image/video)

Output / 1M

Notable extras



Gemini tier (row example)

$0.30

$2.50

Context caching + grounding options



Gemini Flash-style row example

$0.10

$0.40

Very low output cost; good for high volume Tier (example rows from pricing page)

Input / 1M (text/image/video)

Output / 1M

Notable extras



Gemini tier (row example)

$0.30

$2.50

Context caching + grounding options



Gemini Flash-style row example

$0.10

$0.40

Very low output cost; good for high volume Tier (example rows from pricing page)

Input / 1M (text/image/video)

Output / 1M

Notable extras Tier (example rows from pricing page) Tier (example rows from pricing page) Input / 1M (text/image/video) Input / 1M (text/image/video) Output / 1M Output / 1M Notable extras Notable extras Gemini tier (row example)

$0.30

$2.50

Context caching + grounding options Gemini tier (row example) Gemini tier (row example) $0.30 $0.30 $2.50 $2.50 Context caching + grounding options Context caching + grounding options Gemini Flash-style row example

$0.10

$0.40

Very low output cost; good for high volume Gemini Flash-style row example Gemini Flash-style row example $0.10 $0.10 $0.40 $0.40 Very low output cost; good for high volume Very low output cost; good for high volume Gemini’s pricing page also lists: context caching prices, and
grounding with Google Search pricing/limits. context caching prices, and context caching prices grounding with Google Search pricing/limits. grounding with Google Search 3.4 DeepSeek (API) DeepSeek publishes pricing in its API docs and on its pricing page. Model family (per DeepSeek pricing pages)

What to expect



DeepSeek-V3 / “chat” tier

Very low per-token pricing compared to many frontier models



DeepSeek-R1 reasoning tier

Higher than chat tier, still aggressively priced Model family (per DeepSeek pricing pages)

What to expect



DeepSeek-V3 / “chat” tier

Very low per-token pricing compared to many frontier models



DeepSeek-R1 reasoning tier

Higher than chat tier, still aggressively priced Model family (per DeepSeek pricing pages)

What to expect Model family (per DeepSeek pricing pages) Model family (per DeepSeek pricing pages) What to expect What to expect DeepSeek-V3 / “chat” tier

Very low per-token pricing compared to many frontier models DeepSeek-V3 / “chat” tier DeepSeek-V3 / “chat” tier Very low per-token pricing compared to many frontier models Very low per-token pricing compared to many frontier models DeepSeek-R1 reasoning tier

Higher than chat tier, still aggressively priced DeepSeek-R1 reasoning tier DeepSeek-R1 reasoning tier Higher than chat tier, still aggressively priced Higher than chat tier, still aggressively priced 4) Latency: don’t use fake “average seconds” tables Most blog latency tables are either: measured on one day, one region, one payload, then recycled forever, or
pure fiction. measured on one day, one region, one payload, then recycled forever, or pure fiction. Instead, use two metrics you can actually observe: two metrics you can actually observe TTFT (time to first token) — how fast streaming starts
Tokens/sec — how fast output arrives once it starts TTFT (time to first token) — how fast streaming starts TTFT (time to first token) Tokens/sec — how fast output arrives once it starts Tokens/sec 4.1 Practical latency expectations (directional) “Mini/Flash” tiers usually win TTFT and throughput for chat-style workloads.
“Reasoning” tiers typically have slower TTFT and may output more tokens (more thinking), so perceived latency increases.
Long context inputs increase latency everywhere. “Mini/Flash” tiers usually win TTFT and throughput for chat-style workloads. “Reasoning” tiers typically have slower TTFT and may output more tokens (more thinking), so perceived latency increases. Long context inputs increase latency everywhere. everywhere 4.2 How to benchmark for your own product (a 15-minute method) Create a small benchmark script that sends: the same prompt (e.g., 400–800 tokens),
fixed max output (e.g., 300 tokens),
in your target region,
for 30–50 runs. the same prompt (e.g., 400–800 tokens), fixed max output (e.g., 300 tokens), in your target region, for 30–50 runs. Record: p50 / p95 TTFT,
p50 / p95 total time,
tokens/sec. p50 / p95 TTFT, p50 / p95 total time, tokens/sec. Then make the decision with data, not vibes. 5) Compatibility: why “tooling fit” beats raw model quality A model that’s 5% “smarter” but breaks your stack is a net loss. 5.1 Prompt + API surface compatibility (what breaks when you switch models) | Feature | OpenAI | Claude | Gemini | Open-source (self-host) | |---|---|---|---| | Strong “system instruction” control | Yes (explicit system role) | Yes (instructions patterns supported) | Yes | Depends on serving stack | | Tool / function calling | Widely used in ecosystem | Supported via tools patterns (provider-specific) | Supports tools + grounding options | Often “prompt it to emit JSON”, no native tools | | Structured output reliability | Strong with constraints | Strong, especially on long text | Strong with explicit schemas | Varies a lot; needs examples + validators | | Caching / batch primitives | Cached input pricing published | Provider features vary | Context caching explicitly priced  | You implement caching yourself | 5.2 Ecosystem fit (a.k.a. “what do you already use?”) If you live in Google Workspace / Vertex-style workflows, Gemini integration + grounding options can be a natural fit.
If you rely on a broad third-party automation ecosystem, OpenAI + Claude both have mature SDK + tooling coverage (LangChain etc.).
If you need data residency / on-prem, open-source models (Llama/Qwen) let you keep data inside your boundary, but you pay in MLOps. If you live in Google Workspace / Vertex-style workflows, Gemini integration + grounding options can be a natural fit. Google Workspace / Vertex-style workflows If you rely on a broad third-party automation ecosystem, OpenAI + Claude both have mature SDK + tooling coverage (LangChain etc.). If you need data residency / on-prem, open-source models (Llama/Qwen) let you keep data inside your boundary, but you pay in MLOps. data residency / on-prem 6) The decision checklist: pick models like an engineer Step 1 — classify the task High volume / low stakes: tagging, rewrite, FAQ, extraction
Medium stakes: customer support replies, internal reporting
High stakes: legal, finance, security, medical-like domains (be careful) High volume / low stakes: tagging, rewrite, FAQ, extraction High volume / low stakes Medium stakes: customer support replies, internal reporting Medium stakes High stakes: legal, finance, security, medical-like domains (be careful) High stakes Step 2 — decide your stack (the “2–3 model rule”) A common setup: Fast cheap tier for most requests
Strong tier for hard prompts, long context, tricky reasoning
Optional: realtime or deep reasoning tier for specific UX/features Fast cheap tier for most requests Fast cheap tier Strong tier for hard prompts, long context, tricky reasoning Strong tier Optional: realtime or deep reasoning tier for specific UX/features realtime deep reasoning Step 3 — cost control strategy (before you ship) enforce output length limits
cache repeated system/context
batch homogeneous jobs
add escalation rules (don’t send everything to your most expensive model) enforce output length limits cache repeated system/context batch homogeneous jobs add escalation rules (don’t send everything to your most expensive model) 7) A practical comparison table you can paste into a PRD Here’s a short “copy/paste” table for stakeholders. Scenario

Priority

Default pick

Escalate to

Why



Customer support chatbot

Latency + cost

gpt-4o-mini (or Gemini Flash-tier)

gpt-4.1 / Claude higher tier

Cheap 80–90%, escalate only ambiguous cases



Long document synthesis

Context + format stability

Claude tier with strong long-form behaviour

gpt-4.1

Long prompts + structured output



Coding helper in IDE

Tooling + correctness

gpt-4.1 or equivalent

o3 / o1

Deep reasoning for tricky bugs



Privacy-sensitive internal assistant

Data boundary

Self-host Llama/Qwen

Cloud model for non-sensitive output

Keep raw data in-house Scenario

Priority

Default pick

Escalate to

Why



Customer support chatbot

Latency + cost

gpt-4o-mini (or Gemini Flash-tier)

gpt-4.1 / Claude higher tier

Cheap 80–90%, escalate only ambiguous cases



Long document synthesis

Context + format stability

Claude tier with strong long-form behaviour

gpt-4.1

Long prompts + structured output



Coding helper in IDE

Tooling + correctness

gpt-4.1 or equivalent

o3 / o1

Deep reasoning for tricky bugs



Privacy-sensitive internal assistant

Data boundary

Self-host Llama/Qwen

Cloud model for non-sensitive output

Keep raw data in-house Scenario

Priority

Default pick

Escalate to

Why Scenario Scenario Priority Priority Default pick Default pick Escalate to Escalate to Why Why Customer support chatbot

Latency + cost

gpt-4o-mini (or Gemini Flash-tier)

gpt-4.1 / Claude higher tier

Cheap 80–90%, escalate only ambiguous cases Customer support chatbot Customer support chatbot Latency + cost Latency + cost gpt-4o-mini (or Gemini Flash-tier) gpt-4o-mini (or Gemini Flash-tier) gpt-4o-mini gpt-4.1 / Claude higher tier gpt-4.1 / Claude higher tier gpt-4.1 Cheap 80–90%, escalate only ambiguous cases Cheap 80–90%, escalate only ambiguous cases Long document synthesis

Context + format stability

Claude tier with strong long-form behaviour

gpt-4.1

Long prompts + structured output Long document synthesis Long document synthesis Context + format stability Context + format stability Claude tier with strong long-form behaviour Claude tier with strong long-form behaviour gpt-4.1 gpt-4.1 gpt-4.1 Long prompts + structured output Long prompts + structured output Coding helper in IDE

Tooling + correctness

gpt-4.1 or equivalent

o3 / o1

Deep reasoning for tricky bugs Coding helper in IDE Coding helper in IDE Tooling + correctness Tooling + correctness gpt-4.1 or equivalent gpt-4.1 or equivalent gpt-4.1 o3 / o1 o3 / o1 o3 o1 Deep reasoning for tricky bugs Deep reasoning for tricky bugs Privacy-sensitive internal assistant

Data boundary

Self-host Llama/Qwen

Cloud model for non-sensitive output

Keep raw data in-house Privacy-sensitive internal assistant Privacy-sensitive internal assistant Data boundary Data boundary Self-host Llama/Qwen Self-host Llama/Qwen Cloud model for non-sensitive output Cloud model for non-sensitive output Keep raw data in-house Keep raw data in-house Final take “Best model” is not a thing. There’s only best model for this prompt, this latency budget, this cost envelope, and this ecosystem. best model for this prompt, this latency budget, this cost envelope, and this ecosystem If you ship with: a measured benchmark,
a 2–3 model stack,
strict output constraints,
and caching/batching, a measured benchmark, a 2–3 model stack, strict output constraints, and caching/batching, …you’ll outperform teams who chase the newest model every month.

This story contains AI-generated text. The author has used AI either for research, to generate outlines, or write the text itself. 

OpenAI

Choosing an LLM in 2026: The Practical Comparison Table (Specs, Cost, Latency, Compatibility)

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

A Developer's Guide to Merging AI with the Spring Ecosystem

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

The Noonification: White Man (11/26/2022)

The Noonification: The Metaverse is a Sh*tshow (11/2/2022)

100 Days of AI Day 1: From Newsletter to Podcast, Leveraging AI for Audio Transformation

A Developer's Guide to Merging AI with the Spring Ecosystem

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

The Noonification: White Man (11/26/2022)

The Noonification: The Metaverse is a Sh*tshow (11/2/2022)

100 Days of AI Day 1: From Newsletter to Podcast, Leveraging AI for Audio Transformation

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps